<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>If you are a frequent visitor to this site, I hope you will be sad to see that it has fallen into neglect. The neglect is reflected by the log stats too, this week the site has delivered only half as many pageviews as it did the last week of November. Now it could be because of the holidays. I am hoping it is but last year there wasn't such a drastic drop.</p>
            <p>One of my resolutions for the new year that's coming up definitely will be to take better care of this site. I will definitely fix the wiki. It brought many moons ago due to an incompatibility between PHP and Mediawiki. Fixing the wiki is a lot of hard work because mediawiki (sadly) has become bloatware. A new software is needed.</p>
            <p>The site doesn't make any real $ fortunately hosting is sponsored by Rad Inks so it doesn't really matter but <a href="../../megaupload">megaupload</a> drinks up a lot of bandwidth do it's about time that the site starts making some money. A redesign with the adsense ads coming closer to the top of the page will eventually have to be done.</p>
            <p>This blog is static! though it's powered by Movable Type, the MT installation is actually on my local computer and not on this server. I post on my desktop/server and then use rsync to beam it up. The reasons being bandwidth at the local end costing a lot more when the blog first started than it costs now, and a fear of 'comment spam'</p>
            <p>Now that there have been so many posts it's kind of hard to migrate to another blogger and finding the time for the migration is going to be the hardest task of them all.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Neglected

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Finally found a YUM repository that has firefox 1.5. It's a place called <a href="http://www.nrpms.net">nrpms</a>. The yum update went very smoothly, none of the hassles found in the binary download on <a href="https://www.mozilla.com">mozilla.com</a> or the RPMs available all over the place.</p>
            <p>Since all the plug ins have to be reinstalled, I thought this is the best time to update the JVM (in view of the recent security issue). All the extensions have to be updated too, unfortunately quite a few of the extensions do not have updates and cannot be used with firefox.</p>
            <p>Of all the extensions, Live HTTP Headers and open Java console are the most important as far as I am concerned. Hat's off the guys who made them. Sadly both these extensions will not work on firefox 1.5</p>
            <p>Strangely enough when you access a page that contains an applet the browser will automatically pop up a java console. This rather annoying, specially for non technical users. If you close the console and view another applet, the console will not reappear and there is no way to make it reappear. This is rather annoying if you happen to be a developer.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Yum my Firefox

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Setting up and starting apache 2.2 went smoothly. The first hiccup occured when a virtual host was added. The site cannot be browsed. The log file says:</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Apache 2.2 client denied by server configuration

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>For the first time in a long time, a bug fix release of <a href="http://www.radinks.com/upload/dnd.php">Rad Upload</a> had to be made available.</p>
            <p>The previous release, version 2.20 introduced the ability to fire a <a href="http://www.radinks.com/upload/plus/javascript.php">javascript notification </a> when upload completes. It unfortunately introduced a small bug too. Users complained that the max_upload property, which can be used to impose a limit on the maximum size of the upload was being ignored.</p>
            <p>There are two different ways in which parameters can be passed to the applet, first by embedding it in the web page that contains the applet and the second way is to place the parameters in a radupload.properties file. It's the first approach that users had trouble with. Simply moving the max_upload settings to the radupload.properties file did the trick but we made a new release anyway.</p>
            <p>This is the last of the 2.xx versions. The next version of will Rad Upload 3 and we plan to add a loads of new features. The first beta will be available in time for Christmas.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Rad Upload 2.21

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p><a href="https://httpd.apache.org">Apache</a> 2.20 has been released. One of the most exciting features (as far as I am concerned) is large file support. Yes you can now use your favourite webserver to upload and download files greater than 2Gb in size. Naturally the ablity to upload such large files makes me very excited.</p>

            <p>Let's get on the to installation. These is the configuration directive used:

            </p>
            <div style="color:#000033; margin-left:15px">--enable-deflate --enable-rewrite --enable-so --with-mpm=prefork --enable-nonportable-atomics=yes --prefix=/usr/local/apache220</div>

            <p>I want to preserve my 2.0.xx installation so the new version goes into a different folder. It will listen on port 81 for sometime, while port 80 will be listened to by the old version. After some time the roles will be reversed.</p>

            <p>There wasn't any trouble at all with the <em>./configure</em>, <em>make</em> or <em>make install</em> steps. The httpd.conf file has to be edited to reflect the different port number before the server can be started.</p>

            <div style="color:#000033; margin-left:15px">Listen 81</div>

            <p>the start up message seems to have changed too.</p>

            <div style="color:#000033; margin-left:15px">It works!</div>

            <p>Now to install PHP 5.1.1. Since things have gone smoothly upto now, something should fail.</p>

            <div style="color:#330000; margin-left:15px">
                <pre>Warning!  dlname not found in /usr/local/apache220/modules/libphp5.la.
Assuming installing a .so rather than a libtool archive.
chmod 755 /usr/local/apache220/modules/libphp5.so
chmod: cannot access `/usr/local/apache220/modules/libphp5.so': No such file or directory
apxs:Error: Command failed with rc=65536
.
make: *** [install-sapi] Error
</pre>
            </div>

            <p>There's more to life than poking around a PHP engine, so I gave up on it and compiled 4.4.1 with the save configuration directive and had no trouble with the install step either. PHP5 can be used with the 2.0 version of apache.</p>




        </div>
    </div>


</body>

</html>

Apache 2.20

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>After having gone through the thoroughly boring installation steps for Cold Fusion MX 7, now it's time to figure out how to start the server. They seem to have conviniently forgotten to put this information in a readme file or to include it in the getting <a href="https://www.macromedia.com/examples/cfgettingstarted/experience/index_content.cfm">started page on the Macromedia site.</a>. A quick look at the bin/ folder reveals that that the server can be started with the <em>coldfusion start</em> command. (duh obvious).</p>
            <p>The getting started page does not include a hello world example either. I rekoned that the following will do:</p>
            <p>&lt;CFOUTPUT&gt;Hello World&lt;/CFOUTPUT&gt;</p>
            <p>That still leaves me with the task of figuring out where the file should be placed. My installation folder is opt/coldfusionmx7 so I reckoned opt/coldfusionmx7/wwwroot would be a good place. Wrong.</p>
            <p>Type http://localhost:8500/ into the browser, A directory listing will be revealed. It clearly shows the presence of my hello.cfm file. Click on it, and the server written by a bunch of morons will tell you that:</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Getting started with Cold Fusion.

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I am trying out Cold Fusion for the very first time. There was a big fuss many years ago about <a href="https://en.wikipedia.org/wiki/Cold_fusion">cold fusion</a> of Hydrogen. It turned out to be a damp squib. Coldfusion software leaves me a bit cold. The download is a whopping 276Mb. There are complete linux distros that include JVMs and GCC that happen to be smaller.</p>
            <p>You would expect a product with such a gigantic download to have a GUI installer. In the case of Macromedia, the company famous for their UIs, they haev simply overlooked this point.</p>
            <p>A huge portion of that installer seems to be taken up by a JVM, which is more than a bit silly because an update date version of the JVM can be downloaded separately by those who do not have it.</p>
            <p>The EULA seems to take up a fair precentage of those 276 Mega bytes as well. There are so many pages to scroll through, you get the impression that this is an attempt to confuse even the most seasoned lawyer.</p>
            <p>Finally, the installer took me to the next step where it contained about not been able to find C++ compatibility library. It's looking for an old 2.2.5.x while what comes with Fedora Core 4 is 2.3.5.x</p>
            <p>The next step you see another reason why the installer is so bloated; the same installer works for three different editions of Cold Fusion. This is even more lame than old versions of Jbuilder wherethe only difference between the free and the enterprise editions were in the license key.</p>
            <p>Just when you get fed up with the whole thing it tells you a joke, just to lighten your day. <em>After</em> presenting a list of choices in a (text only ) menu, you are asked to 'Please select one of the following options'.</p>
            <p>After answering another half a dozen questions the installer finally copied the files to the HD. It takes up all of 400 MB. Windows users may think nothing of it but on linux we don't use bloatware.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

ColdFusion leaves me cold

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Downloaded firefox 1.5, like the last time I tried to <a href="545.html">install firefox</a> things refused to go smoothly. Interesting that the download site has changed to <a href="https://www.mozilla.com">mozilla.com</a>. What hasn't changed is the insistence that they do a browser detection and try to force a file down your throat.</p>
            <p>The firefox-1.5.tar.gz file that was delivered does not contain an installer, you just expand it and run. But run it will not.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Firefox One and a Half

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>The last thing that java comunity need now is a serious java bug that would get widely reported on security sites. Unfortunately it has happened and this bug seems to effect all versions relased over last 5 years or so.</p>
            <p>The timing couldn't have been worse. Java has been gaining ground. The last serious flaw has been long forgotten by the public and even by the community. In the meantime java has taken over as the #1 technology at sourceforge. Products like Azureus have done a lot for the popularity of the java. I like to think that at Rad Inks we have played a role too with our applets.</p>
            <p>So in the middle of this we are now told that there is a bug in the sandbox that results in unsigned applets being able to access privileged resources. An exploit has not been published, but there may have been one published in underground sites.</p>
            <p>See the</p>
            <p><a href="http://www.javalobby.org/java/forums/t55086.html">Java Lobby discussion</a></p>
            <p><a href="https://secunia.com/advisories/17748/">Secunia report</a></p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

A java bug.

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Just tried two of googles services, one made my head spin, the other made me wonder if google is loosing it's marbles.</p>
            <p>Urchin Web Analytics, now called Google Analytics used to be a $199 per month service. Just recently they started to offer it free of charge to adwords users. Since Rad Inks spends quite a bit of money on adwords, I decided to give urchin ago.</p>
            <p>It was only recently that we enabled conversion tracking, resisting the urge to add that javascript which created the google cookie. Afterwards tracking conversions for a while one of two conclusions can be reached. A)Spending money on adwords is a waste. Our cost per sale is far in excess of our revenue per sale. B) Conversion tracking is woefully inaccurate. When we cut down on ad spending our sales drop.</p>
            <p>It was in this backdrop that analytics was tried out. Even though the analytics JS is plainly visible on the pages on our site, google insists it's not there. I must have clicked the check status button a dozen times. What's really funny is that the log file does not show any identifiable google robot attempting to fetch anything from the site when the check status button is clicked. Somewhat suprisingly, a couple of pages in the analytics help section throw up 404s.</p>
            <p>The other google service that made my head spin was <a href="http://base.google.com">google base</a>. It sounds like a cross between the open directory, Ebay, and a classified system. Not suprisingly google has applied for some patents on classified systems.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Google Urchin on Home Base?

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p><a href="http://freshmeat.net">Freshmeat</a>, being such a large site, appears at the top for many search queries that revolve around software. Until today the description (the extract) that follows the link was a true extract of the webpage concerned.</p>
            <p>All that has changed. If you type in Rad Upload into google, this what you see:</p>
            <p></p>
            <div style="margin-left:15px; margin-right:15px; color:#000033">
                <p></p>
                <p><a href="http://freshmeat.net/projects/radupload/">freshmeat.net: Project details for Rad Upload</a><br></p>
                <p>freshmeat maintains the Web's largest index of Unix and cross-platform open source software. Thousands of applications are meticulously cataloged in the ...</p>
                <p></p>
            </div>
            <p></p>
            <p>The link is correct but the extract is not correct. If you type in the name of any other project listed on freshmeat you will see the same description.</p>
            <p>This can only mean one of two things: a) Freshmet has started to cloak in a massive way - Which is very unlikely. b) Google has a serious bug. That is also unlikely. So what are we left with? Speculation.</p>
            <p></p>

        </div>
    </div>


</body>

</html>

Google bug?

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>We should not make disparaging remarks about ASP.NET developers. They are the less fortunate in society. We should sympathize with this disadvantaged community who have never had the opportunity to write code in PHP or Java.</p>
            <p>Take for example the case of the include or require calls in PHP. These language constructs (They are not really functions) can be used to include another file in the response returned by the script on execution. The files themselves can be dynamic or static.</p>
            <p>Both PHP and Java provide at least four different ways in which another file can be included. That makes is very easy to create template driven websites.</p>
            <p>The misguided fools who only develop with VB.NET or C# have been brainwashed into believing that not having such a feature is a good thing. They have been taught at .NET has a superior architecture. Anything that is a copy of Java architecture whould be a good thing, but they have not copied everything. That is they have missed a bit.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

ASP.NET and Include.

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Grr, who designed the structure for the ASP.NET documentation? I have been scratching my head for nearly an hour trying to figure out how to place a message in the apache log file.</p>
            <p>When you search you come up with 100 line solutions written by M$'s Most Valuable Pricks that show you how to write to windows event logs, send errors as email messages and send them as messages in a bottle. All of them seem to ignore this simple trivial one line solution.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

ASP.NET logging to Apache error_log

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I first ran into the .NET platform in 2000, despite that I am happy to say I have never written a line of .NET code until now. My previous close encounter was purely to do with the fact that I was waging a lone battle for the honour of open source against an army of MCPs at a Microsoft certified solution provider. I read up a lot on .NET with out actually writing a single line of code.</p>
            <p>I am happy to say that many of my former collegues are now die hard PHP fans, even though they may not exactly be linux fans. One of them happens to have about a dozen microsoft qualifications including MCSE and MCDBA. But I digress.</p>
            <p>On friday, I installed the <a href="http://www.go-mono.com/">Mono</a> cause I was bored. Unfrotunately the all in one installer that I downloaded does not seem to include, mod_mono which is required if you wish to write web apps with .NET on linux. Mod_mono as you rightly suspect is the apache module that enables it to run ASP.NET scripts. (are they called scripts? who cares?)</p>
            <p>Mod_mono had to be downloaded and compiled by hand. Though all the sample code included in the download did run, my own hello world didn't work at all and it took me a while to realize the errors of my ways.</p>
            <p>Well, I haven't really realized the errors of my ways but I will pretend that I have. Which is fair enough because .NET is all about pretending. It pretends that it wasn't influenced by java to start off with. Someone please explain to me why .NET pretends to be even driven. HTTP simply doesn't work that way, but try explaining that to the idiots staring at a blue screen of death or formatting their hard disks because another virus got through IE.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

ASP.NET and Mono.

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I have been experimenting with HTTP push. It doesn't work too well with IE so I had a look at the over hyped AJAX to see if it is any good. On the surface it is.</p>
            <p>You can listen to the readyStateChangeEvent and get notification when new data arrives. Obviously you will have to open your XMLHttpRequest object in asynchronous mode for this to work. However there is a catch you cannot get hold of the responseXML object. It gives you a null while the responseText object is populated.</p>
            <p>If you want to get hold of a DOM object you need to create a DOMParser and get it to parse the responseText. None of these are things that will give you nightmares what will give you a nightmare is what happens when you try to make another XMLHttpRequest while the PUSH is happening in the background. The browser freezes up.</p>
            <p>Does that mean that there can only be one XMLHttpObject instance per window?</p>
            <p>The mozilla javascript documentation certainly leaves much to be desired.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

How asynchronous is AJAX

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>It's election time in Sri Lanka. But as you know politicians don't really deserve your vote. It's much better to spend your time voting for a more worth while cause; such as my listings on <a href="https://www.hotscripts.com/">Hotscripts</a></p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Cast Your Vote

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>It's been an year since the last release on the Mega Upload PHP branch. The project itself is nearly two and a half years old and now used at over <a href="https://sourceforge.net/project/stats/detail.php?group_id=87646&amp;ugn=megaupload&amp;type=prdownload&amp;mode=alltime&amp;package_id=0"> a 100,000 thousands websites.</a></p>
            <p>Over the past year, there have been few contributions coming in, unfortunately due to various other commitments I never had the chance to release any of them. Most of the contribs were attempts to rewrite the progress monitor with AJAX.</p>
            <p>I am uncertain about the advantage of doing so. True that using AJAX would mean that the progress monitor does not have to refresh frequently, but what advantages is there in it? In reality XMLHttpRequest does result in another request being sent to the server. So the load and the number of hits on the server remain unchanged.</p>
            <p>It is possible that I may change my mind about AJAX. I might just find the time to make another release of megaupload with an AJAX monitor. Right now the effort is on building a C version of Mega Upload, this is something that has been continously put off for over 2 years.</p>
            <p>Currently <a href="../../megaupload">Mega upload</a> works by getting perl to accept the file upload and write information to the disk, which is then read by the progress monitor. After the upload completes handling is passed back to your PHP script.</p>
            <p>The only catch is that perl isn't that great when it comes to reading streaming data (PHP on the other hand is terrible). C would be ideal but parsing a multipart form post with C isn't something for the faint hearted or lazy.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Mega Upload 2

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Is it possible to do a file upload with javascript? Short answer: no. Is it possible to trap file upload events with javascript? the short answer is again no. But there is always a way.</p>
            <p>Fortunately in software it's very dangerous to say something is impossible. (People told us <a href="http://www.radinks.com/sftp/">Rad SFTP</a> is impossible to build but we built it anyway).</p>
            <p>We have just added a new feature to <a href="http://www.radinks.com/upload/">Rad Upload</a>, That makes it very easy to recieve javascript notifications when an upload is completed. In fact it's ridiculously easy. All you need to do is uncomment a single line in the configuration file. Then once the file transfer is completed the applet will call a javascript function named uploadCompleted(). The response returned by the server is passed through as a parameter.</p>
            <p>In addition to the callback, Rad Upload exposes two methods that can be invoked by javascripts. These can be used to determine if the last upload was successfull, failed or was cancelled. Another method allows you to retrieve the last response returned by the server.</p>
            <p>Now to get back to the original question. Is it possible to do a javascript upload? One solution is to call a method in Rad Upload from your javascript and ask the applet to start uploading. However the method to start the upload has been intentionally kept hidden. No prizes for guessing why.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Javascript file upload

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>The attempt the other day to <a href="829.html">design a GUI with Netbeans 5</a> didn't meet with much success. I assumed the reason for this was because I was working on a design that was initiated with the older 4.1 version. Today I made a fresh attempt starting from scratch and met with a certain amount of success.</p>
            <p><a href="https://www.netbeans.org">Netbeans</a> happily allowed me to drag and drop components as I pleased and their placements was at the most natural location (4.1 insisted on placing them at the most un-natural location). It still wouldn't allow you to rename the tabs in a tabbed pane but nobody is perfect. However after some time netbeans started to act up again. The same exception that was seen yesterday showed it's face when a component was resized. The best way to design GUIs is by hand with a GridBagLayout!</p>
            <p>Went on to other things but I found that whatever source file that I opened the indentation was wrong. No amount of juggling with the indentation settings would put that right. Worse still netbeans now insists on placing the brackets at the wrong positions. If that setting can be changed it's cleverly hidden.</p>
            <p>Time to uninstall the 5.0 Beta and go back to the 4.1 version that seems to work well.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Netbeans 5 Beta (Part II)

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>A couple of days ago I wrote about how Rad SFTP was highly rated at <a href="http://www.jars.com">jars.com</a>. I also wrote about how <a href="http://www.radinks.com/upload/">Rad Upload</a>. a product that was also submitted to jars.com around two years ago didn't get rated until just a few days ago.</p>
            <p>Almost as soon as the post was published there is another mail from jars.com they have rated <a href="http://www.radinks.com/sftp/">Rad SFTP</a>. Again. This time the rating is much lower no mention of stars.</p>
            <p>The reviewers are said to be volunteers so they don't get paid for reviewing the resources, so can anybody tell me what's happening?</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Jars.com Double Take

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Thawte has a very poor idea about service and customer support, so when they asked if I would like to participate in a survey I jumped at it. At the outset the survey site mentioned that it could take about 10 minutes to completed. I decided to go ahead with that just so I could let off a little bit of steam.</p>
            <p>Unfortunately it seems Thawte and/or the company they have hired to do the survey are <a href="796.html">clueless</a> about the products that thawte have on offer. I have never obtained an SSL certificate from thawte and never will. There are other organizations that offer SSL certificate and give a better service.</p>
            <p>I only ever ordered <a href="729.html">code signing certificates</a> from them. Yet on the survery question after question is about SSL certificates. No where do they mention any other kinds of certificates. The moron who designed the survey only needs to visit the thawte website to find what products they sell. Or perhaps the moron is incapable of reading.</p>
            <p>I jumped to the latter conclusion and abandoned the survey half way through. Hopefully the comments will be passed onto someone who can read.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Another Joke By thawte

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Two years ago we submitted the <a href="http://www.radinks.com/sftp/">Rad SFTP</a> applet at jars.com. They liked it. They gave a very high rating. Highly rated applets at at <a href="http://www.jars.com">jars.com</a> used to be featued on <a href="http://java.sun.com">java.sun.com</a>. As a result Rad SFTP was featured there too.</p>
            <p>Then we submitted <a href="http://www.radinks.com/upload/dnd.php">Rad Upload</a>, that was also about two years ago. We never heard back from jars.com since then. That is we didn't hear from them until yesterday. Two years after the submission was made they have written to inform us that the product has recieved 875 points. What percentage that belongs to there is no mention.</p>
            <p>So what's happening at jars.com have they been doing a Rip Van Winkle?</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

What's up at jars.com?

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>My new broker is part of the Colombo Direct Access (CDAX) system and I downloaded CDAX 5.4 with much enthusiasm only to find that they have sold themselves short with the installer. Why people choose to create binary installers with native code for java applications is something that I will never figure out.</p>
            <p>Here is this reason that CDAX is distributd in the form of a binary: the download includes the JVM. Sheesh you would think a linux user would know how to install java on his own. I have three different versions of the JVM running side by side why do I need a fourth?</p>
            <p>It wouldn't really matter if the installer works. It does not. It dies lamerly with the suggestion that that errors.log file be looked at for details. It says:</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

CDAX on Linux

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>The author of <a href="http://proguard.sourceforge.net/">Proguard</a>, in the true spirit of open source software development, has links to other alternative obfuscators/optimizers on his site. I have already seen some of them because they are also included in the list at <a href="http://mindprod.com">mindprod</a>. There are exceptions and jode is one of them.</p>
            <p><a href="822.html">Proguard as we have already seen is ahead of the rest</a>. On first glance jode seems to be reasonably close in many aspects the difference though is in the documentation.</p>
            <p>The last obfuscator that was looked at is <a href="http://www.nq4.de/">JoGa</a>. It has a nice graphical user interface and seemed to work well, but when processing a large project it encountered a stack overflow. This can probably be overcome by tweaking the JVM settings.</p>
            <p>Run back again to Proguard</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

More on Obfuscators

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Once upon a long ago <a href="180.html">I had a Sinclair ZX spectrum</a> with all of 16 Kilo Bytes of RAM. Today finding a file smaller than 16KB on my 120GB hard disk is a tough ask.</p>
            <p>What set me off on this bout of reminiscing was my acquistion of a DVD writer. I used to have a CD ROM before many other geeks and the same pattern was followed for a CD writer. But when it comes to reading or writing DVDs I have been lagging behind. So when I went to the retailer to pick up a new hard drive (<a href="000840.html">after the annual crash</a>), decided to pick up a DVD writer as well.</p>
            <p>It's double layer so it can burn upto 8 GB on a single disk. Less than 5 years ago I had a hard disk of the same size. Now I have three hard disks totaling 200GB. My first hard drive on the other hand was a mere 10MB. My first computer didn't have any disks at all. You had to save all the data on a cassete tape (rembers those audio cassete tapes?)</p>
            <p>The 8086 I had long ago (the first member of the Pentium series) had a 360kb floppy drive. In fact it had two of them. The first was for the program disk the second for the data disk. It had all of 640Kb of memory. Geeks with only one</p>
            <p>drive used to create a RAM disk for holding the program disk.</p>
            <p>The next machine I had if I remember correctly was a 386 with 8MB of ram and the 10 MB hard drive. From that I graduated to a Pentium 133Mhz with 180MB and 16 MB ram which was later upgraded to 32MB. To this machine I later added a CDROM and an 2GB hard disk.</p>
            <p>From there it was to a K6-II followed by a K6-III the memory jumped all the way upto 128MB and the hard drive went upto 8GB. Then I started working and picked up an AMD athlon 700MHz, Memory doubled to 256MB and in between I picked up two drives one of 10 GB and other the 20GB. The 20GB was retired only recently it was serving as a third drive purely for backup purposes.</p>
            <p>The next machine was a total disaster an Athlon XP with a <a href="562.html">totally messed up mother board</a>. It killed off one of my hard drives too. From there I moved to the current machine which is a P4 2Ghz. In I upgraded the RAM to 1GB. That's nearly 1500 times as much as I used to have on my 8086.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

DVD and Spectrums

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I must have installed PHP with XSLT support a hundred times. I have seen this error reported thousands of times in various newsgroups and message boards. I have seen it only once and that was about five years back. But I am seeing it now.</p>
            <p>The <a href="000840.html">annual hard disk</a> crash meant having to install all kinds of software all over again. PHP is one of them. First I chose php-5.10 RC it didn't compile at all with the configuration settings I chose. Since GCC 4 still has many incompatibility I update my compiler (and it's dependencies) to no avail. (Fedora users are really guinea pigs for RHEL)</p>
            <p>Rather than waste time on 5.10, I switched to 5.04 that too refused to compile at first but once I discarded the xml-rpc option it compiled fine. In reality it should have choked on the <em>--enable-xslt' '--with-xslt-sablot'</em> because after the restart they do not show up with a call to phpinfo() and a call to xslt_create() now fails with the error reported in the title.</p>
            <p>Next I tried downloading and installing <a href="http://www.gingerall.com/charlie/ga/xml/d_sab.xml?s=org">Sablotron</a> from the source. That had no effect whatso ever.</p>
            <p>In the end, I decided to switch back to 4.4.0 and that works like a charm.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Fatal error: Call to undefined function xslt_create()

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Last year around about this time I had a hard disk crash and a <a href="562.html">mother board stew</a>. So I wasn't at all suprised when one of the three hard disks plugged into my machine decided to go on strike.</p>
            <p>Well it wasn't really the hard disk's fault or my fault. It's the UPS's fault. The UPS of course would point the fingure back at me and say it's my fault for letting the battery run out. Anyway as often happens in this part of the world; the lights failed and when the came back on less than a minute later the machine refused to boot.</p>
            <p><a href="https://www.gnu.org/software/grub/">Grub</a> got stuck attempting to load a 1.5 stage? WTF is the 1.5 stage? anyway the system always gave an error5 and refused to go any further. The usual remedy of booting off an installer CD and going into rescue mode to run grub-install didn't work. What's really annoying is that all the partitions on the disk and the data contained in them appear to be intact. It's just sector zero that's causing trouble.</p>
            <p>Next tried to run the installer in update mode (<a href="794.html">my system is Fedora Core 4</a>) after going through the entire process the $^&amp;<a href="../../cdn-cgi/l/email-protection" class="__cf_email__" data-cfemail="9cbddc">[email&nbsp;protected]</a> installer says boot loader was <em>not</em> updated because no kernel modules were updated.</p>
            <p>After playing around for quite a bit more including booting with <a href="http://knoppix.com/">knoppix</a> to run grub-instal again and trying various remedies mentioned on various forums I decided to try out a fresh install. Unfortunately that didn't work either. The same error message kept appearing even after a fresh installation of FC 4.</p>
            <p>Finally I decided to throw in the towel and buy a new hard disk. In the software business time really is money. I shouldn't have spent so many hours trying to get to the bottom of this disk failure. It's only 40 GB disk anyway. So went into a computer shop and bought a brand new 120GB hard disk and while I was at it digged a bit deeper into the pockets to pick up a DVD writer as well.</p>
            <p>All this happened on saturday and I am still playing catch up.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Annual Hard Disk Crash

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>The search results on this site used to be produced by <a href="http://aspseek.org/">ASPSeek</a>. ASPSeek is no longer in development and I haven't used the alternative mnogosearch in quite a while. For the uninitated ASPSeek has nothing to do with the Active Server Pages by the M$ company. It's a fork of the mnogosearch project. Both these used to be quite popular not so long ago.</p>
            <p>I am reluctanct to start using <a href="http://search.mnogo.ru/">mnogosearch</a> (a project I have not used since switching to aspseek). I would need to relearn things. The configuration syntax etc for mnogosearch is very similar to that of ASPseek (remember the latter is a fork). The truth is I have been so completely out of touch with both systems that I would need to relearn everything from scratch.</p>
            <p>If you are learning, learn something new. So the site's search pages are now powered by <a href="https://code.google.com/">Google</a>. It turned out to be redicolously easy to implement.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Google API and Site Search

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p><a href="https://lg3d.dev.java.net/">Looking Glass</a> is a very interesting idea by a developer at Sun. I can't wait till it's mature enough for day to day usage, so tried the live cd.</p>
            <p>Tried to start Looking Glass as a guest of Qemu. Unfortunately it didn't meet with much success. The system would boot, it would go into what appears to be the init phrase and a the end instead of bringing up a login screen or starting X it just says hit enter to reboot.</p>
            <p>One might think the X server is running on another virtual screen but holding down the alt key and pressing all the functions keys had no effect.</p>
            <p>The next step was to put the CD into the drive and boot up. Running, Looking Glass doesn't like the look of vector. Unfortunately the result were no different from running under <a href="http://fabrice.bellard.free.fr/qemu/">QEMU</a>. Perhaps it doesn't like my video card, but my video card is an NVidia and this distro is supposedly tilted against NVidia.</p>
            <p>Took out the CD, and tried it on a different machine (a notebook) viola it starts up ok. The start up screen has what I think is probably a desktop, it looks like the <a href="https://lg3d.dev.java.net/branding/images/lg3d-core/lg_snapshot.png">'official snapshot'</a> but it doesn't take up the full screen space. It also has an 'x' in the tool bar to close it. Clicked on it and the whole thing crashed.</p>
            <p>Enough of short cuts, let's download the developer's edition and give it a go.</p>

        </div>
    </div>


</body>

</html>

Looking Glass

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>After the dispointment with the <a href="829.html">GUI designer in Netbeans 4.1</a>, I decided to try out the 5.0 beta, which is supposed to have a different GUI designer. I decided to be adventurous based the installation on java 6 rather than on java 5. It wasn't a wise decision.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

NetBeans 5 Beta

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Have been using <a href="https://netbeans.org">Netbeans</a> for quite some time with out using it's GUI designer. My opinion is that <a href="774.html">Netbeans is better than many of the other IDEs </a>I have had the opportunity to use. But there is certainly room for improvement. The GUI designer is one of them.</p>
            <p>I was quite shocked to learn that the code generated by the Form Designer (as netbeans calls it) cannot be edited by hand. Obviously you can open the source in your text editor and make changes, but you cannot edit it from with in the IDE itself. That sounds like something you might find in an IDE by that Redmond corporation. The behaviour is most Unbecoming of an open source IDE.</p>
            <p>The design that I chose to create was a Tabbed Pane. Suprise , suprise, you cannot change the names for the tabs!</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Netbeans GUI Designer

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>It's rare for linux freaks like myself to defend Microsoft, but Microsoft's standing firm against the Eolas patent law suite does need to be comended. Should Eolos have their way in this patent battle they will then target their guns on Firefox and other browsers.</p>
            <p>When the dust settles, even if M$ ends up paying millions of $ to the plaintiff they will not be able to do away with the plug in mechanism. If by some chance they start pretending that plug ins never existed millions of websites that use Java, Flash, Real and other media types will be broken over night.</p>
            <p>In the end this issue will hopefully expose the evil known as software patents. Microsoft themselves are quite keen to patent various technologies but hopefull they will see the error of their ways.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Eolas and Patent Folly

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Ask me "what is the worst software you have seen?", and I wouldn't have an answer for you. I have seen thousands of really badly written software but I don't even give them a second glance.</p>
            <p>Ask me "What is the worst software you have used?" and I would say it is my stock brokers online trading system. There are dozens of fundamental flows with this application named Atrad.</p>
            <p>One would think that a team of software architects would debate for weeks before coming up with a good design for an online share trading system. It's obvious that an architect worth his salt has not been involved in the development of Atrade.</p>
            <p>The software actually is nothing more than a frontend for the stock brokers database. Stunning as it may seems, Atrad operates by opening a connection to the oracle database on the broker's server. If you disregard the security implications for the users of the system, can you disregard the security implications for the other clients of the brokerage firm?</p>
            <p>You might think that an online trading system would be real time. Not if you are a Sri Lanka. Atrade and the alternative system CDAX used by some other brokers are both written in Java. But what's really amazing is that despite being a Java application, Atrad is available only for windows!! Why on earth write it in java in the first place? why not a real time technology more suited for windows?</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Online Share Trading

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Many users of SLT ADSL complain about the connections slowing down to a crawl and even timeouts when browsing. I also ran into the same situation and tracked it down to a problem with Sri Lanka Telecom's DNS servers.</p>
            <p>Two of their servers '203.115.0.1' and '203.115.0.18' sometimes stop responding completely and you cannot even ping them. The solution then is to use an external DNS server. Many of the other ISP in Sri Lanka have forgotten to block ports on the DNS servers and virtually anyone can use their DNS servers.</p>
            <p>You shouldn't be suprised by this lack of attention to details, quite often you see so called sysadmins from Sri Lankan ISPs/Telecos asking elementary questions in mailing lists.</p>
            <p>A more sophisticated solution than using a third party DNS server is to set up your own. <a href="https://cr.yp.to/daemontools/install.html">daemontools</a> obviously would be the first choice. When setting up however, I ran into a small problem.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

SLT ADSL, DNS and Daemontools

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Had the indepth look at <a href="https://wordpress.org">Wordpress</a> as I threatened to do the <a href="821.html">other day</a>. One of the first things I learnt is that wordpress is actually a fork of b2evolution. I had read that <a href="http://b2evolution.net/">b2evolution</a> was a good product but did not get to study it because of the problem with it's authentication systems.</p>
            <p>Unfortunately wordpress seems to carry this same login system and looking into my cookie jar I found that the username and password is stored in cookie. This is very lame and ridiculous. It makes an otherwise very good software look like it has been written by a rank amature. Of course it's perfectly OK to use cookies, what's not OK is to save the password in a cookie.</p>
            <p>Looking into other issues; it's refreshing to note that the templating system of Wordpress is pure PHP. Smarty and other <a href="../../wiki/Case_against_templates">lame templating engines</a> are no where to be found. They call these templates themse and there are some themes available for download on the wordpress site. What's suprising is that they are so few in number.</p>
            <p>The admin console of this blogger is superior to any other that I have seen and that includes <a href="http://www.sixapart.com">movable type</a>. Wordpress also appears to have a decent comment spam filtering system. Feeds, pings trackbacks, you name it wordpress has it.</p>
            <p>What wordpress doesn't have is the ability to export blog entries into static html. Opinion is devided into the usefullness of this but consider this: the only difference in the URL for different posts is in the query string. That means your posts are not going to get rank very high by google.</p>
            <p>You will point out that it's easy to overcome this with URL Rewrites. If you don't mind having a 43 line .htaccess file that certainly is a possibilty. Don't forget that having lenghty .htaccess files would mean an extra load on the server. Add to that the CPU cycles eaten up by the PHP engine to spit out HTML that could easily have been static.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Wordpress

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Because of the <a href="818.html">incompatibilities between proguard 3.3 and Java 6</a>, I decided to have a look at other obfuscators/optimizers. The authors of the first one I tried, <a href="http://www.retrologic.com/retroguard-download.html">Retro Guard</a> seem to be clueless about the GPL.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Java Obfuscators.

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>A couple of days ago this serial blogger looked at a few blogging tools and found they didn't come up to the standards set by <a href="https://movabletype.com/">Movabletype</a>. Today it's the turn of a few more victims starting from <a href="http://b2evolution.net/">b2evolution</a></p>
            <p>With B2evelution, the reliance on cookies is just too much. The login system failed to work for me and looking through their archives I found that I am not the only one. Each time you click on something you are asked to login again. This prompted me to take a close look at the code and it appears that they may even save the passwords in a cookie!</p>
            <p>Now they may argue that they are saving the password as an MD5, that doesn't matter, you can still create a fake cookie with that.</p>
            <p>They are few more blogger on the same hotscripts page but none of them have a product description to wet the appetite, so they were not even considered. The next page was pretty similar, one had this description:</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Serial Blogger

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Class verification errors were the order of the day when I first started using Java 5. With java six it's something else:</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

runnning with Mustang

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I am at the moment very happy with Movable Type, but it doesn't do any harm to look at the alternatives to see if something else has caught up or overtaken MT while i wasn't looking. The best way to dig up software such as this is with hotscripts.</p>
            <p>The first non commercial entry in the <a href="https://hotscripts.com/PHP/Scripts_and_Programs/Blog/index.html">PHP blog scripts</a> category was <a href="http://mywebland.com/download.php?id=19">myBloggie</a> and the first thing you see on visting the download page is a security fix. On to the second one: <a href="http://boastology.com/pages/download.php">BoastMachine</a>, the download doesn't start (servers seems to be swapped to death).</p>
            <p>The third one, <a href="http://www.webfooted.net/wfblog.php">Web Footed</a> has only an RAR file for download. Who uses RAR? On the fourth attempt, I came across a site that works. It's called <a href="http://www.plogworld.net/">plog</a></p>
            <p>Plog uses smarty. I hate smarty and all other templating engines for PHP. But that would not matter if it plog can generate static pages as movable type can. Hey but wait a minute where is INSTALL.txt? It's wikified and <a href="http://wiki.plogworld.net/index.html/PLog_1.0/Installation">online</a> and quite good. But they should still include an INSTALL.txt</p>
            <p>Apparently installation is only possible using the wizard. That's not so good. Then again the installation was straight forward. Unfortunately it does not look like you can generate static pages with plog.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Looking for a new blogger

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Just downloaded a snapshot of java 6.0 naturally the first thing is to see if <a href="http://www.radinks.com/upload/">Rad Upload</a> works well with it. It does.</p>
            <p>The only noticable difference is in the security certificate that pops up when the applet starts. The new version conveys a better sense of security and it's unlikely to scare off paranoid users.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Java Mustang

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Did a bit of experimenting with <a href="https://www.google.com/webmasters/sitemaps/">google sitemaps</a>. It left me wondering if the effort was worth it or if it is even counter productive.</p>
            <p>This was for a new website. A few days after the site's contents were first upload, the home page was submitted with the standard <a href="https://www.google.com/addurl/?continue=/addurl">form</a>. A few days later the home page and only the home page was indexed. The problem of course is that the site does not have enough incoming links yet. So it was decided to try out sitemaps.</p>
            <p>The sitemap was generated with the python script (one more reason for me to learn python). Then submitted to the sitemaps feature. So far so good but the next day when I checked the status I find that the sitemap is reported as being missing. Hello?</p>
            <p>Found the reason, when synchronizing from the development server to the live server the sitemap file get's deleted (since it only exists on the live server). So the sitemap file was downloaded and a copy was placed in the local server, this time google picked it up with out any problem.</p>
            <p>Now the bad news, a search for the site reveals that the home page which had previously been indexed has now been removed! it's place is taken up by a list of url but none of the url have been indexed so they will not show up in a search.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Google Sitemaps.

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>This is a gloat/boast.</p>
            <p>There are couple of products similar to <a href="http://www.radinks.com/upload/">Rad Upload</a> out there. They can be tagged as similar becuase they are HTTP file uploaders and they can be used to upload multiple files there the similarity ends.</p>
            <p>But these clients do have lot common with each other. They are all playing catch up with Rad Upload. I see that some of them now offer the ability to resize images before upload. One company that has just caught up with us this way has gone to extra ordinary lengths to gain publicity. They have posted a message to a thread that is more than an year old hoping to gain page rank.</p>
            <p>Another common feature is that they are all much much larger in size than Rad Upload. You might say that a 100 kb difference hardly matters given the speed of modern internet connections. The truth is the time taken to download an applet is insignificant when you compare with the time taken to start an applet.</p>
            <p>A 100Kb applet will usually take more than twice as long as a 50kb applet to start up. And when it comes to size Rad Upload is the clear winner.</p>
            <p>In terms of pricing, Rad Upload is once again ahead. While a couple of other products are approximately the same price as Rad Upload, one is nearly four times as high but has a lot less in terms of features.</p>
            <p>All these companies are playing catch up are going to have to put in an extra burst of speed because we have a lot of new features planned. In the meantime, I would like to see what they make of <a href="http://upload.thinfile.com">Thin Upload</a></p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Playing catch up

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Another software that stopped working after <a href="794.html">upgrading to Fedora Core 4</a> was Subversion. I have two installations of <a href="http://subversion.tigris.org">subversion</a> in fact. One uses BDB and the other users FSFS. Since upgrading all my work has been with the FSFS installation and the first time I tried using the BDB version I was in for a nasty suprise.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Subversion on FC4

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p><a href="https://www.google.com/">Google</a> has often been a favourite whipping boy for privacy advocates. It's refreshing to see google moving in the right direction to stop these critics.</p>
            <p>I tried out the recently released official google toolbar for firefox. (An unofficial tool bar has long been available). It is available in the form of an extension and you need to restart the browser once the extension is installed.</p>
            <p>On restarting I was pleasently suprised to see this:</p>
            <p><img align="center" src="http://images.raditha.com/blog/images/toolbar.jpg"></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Google Toolbar for linux

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Trying to debug yesterday's <em>The connection to www.raditha.com terminated unexpectedly</em> error with Apache 2 SSL. Tried using the openssl client (<em>openssl s_client -connect raditha.com:443 -state -debug</em>). It produces the following result; As always there are lots of pundits with strong opinions, weak reasoning and no real answers.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

SSL23_GET_SERVER_HELLO:unknown protocol

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Having had <a href="801.html">yet another run in with Apache</a> SSL (mod_ssl) and being very disapointed with the service of Thawte, I decided to have a go at setting up a certificate authority.</p>
            <p>Setting up a CA doesn't really serve any purpose because no one is going to bother to install my cacert into their browser but here isn't always a purpose to everything that a geek does. Besides a friend of mine, a cryptograpy specialist had thrown me challenge to set one up (this is a separate development).</p>
            <p>As usual being <a href="http://fedora.redhat.com">Fedora</a> user I was at a disadvantage. The <a href="https://www.openssl.org">openssl</a> installation on the system by Red Face does not include the CA.pl (or CA.sh) script. And it does not include the sign.sh script. Not having these scripts means you need to do a hell of a lot of typing. Fortunately, I could dig up a copy of <a href="http://www2.goldfisch.at/knowledge/142">sign.sh</a> with a little googling and decided to make do with out CA.pl</p>
            <p>Even without CA.pl creating a root certicate is straight forward and howto do so is well documented in the <a href="https://httpd.apache.org">apache</a> SSL guide. So I should be signing off now but as often happens what works on the local server does not work on the production server. When ever the https url is accessed all you see is that famous message:</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Setting up a certificate authority

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p><a href="http://www.radinks.com/upload/">Rad Upload</a> keeps getting better with each release. But the last release was exactly two months ago. So it's just the right time to come out with version 2.11.</p>
            <p><a href="http://www.radinks.com/upload/changes.php">What's new</a> in the new version? there is a boring bug fix. What's exciting is the improvement to the file selection dialog. The main thrust of Rad Upload's marketing campaign when it was first launched was that it is a drag and drop file uploader with a progress monitor.</p>
            <p>Then we added the copy and paste support so that it would work well on linux as well. Many linux desktops didn't allow drag and drop file upload at the time (but now they do).</p>
            <p><a href="http://www.radinks.com/upload/plus/browse.php"><img border="0" src="http://www2.radinks.com/images/upload/browse-t.jpg" style="float:right; clear:left; margin-left:5px"></a></p>
            <p>With version 2.0 came the browse button (and many more things besides). Apparently some users still wanted to browse button (who are we to argue). Unlike the browse button on traditional HTML form based file uploaders, this one allows you to select multiple files and folders at once.</p>
            <p>The most recent version just released allows a preview of images chosen for upload.</p>
            <p></p>

        </div>
    </div>


</body>

</html>

An even better file uploader

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>My last run in with <a href="405.html">Apache SSL wasn't very pleasent</a>. But that was with apache.3 where SSL support is not built in. Version 2.0 has built in support thanks to the export restrictions been done away with.</p>
            <p>So it's time to have a go at installing apache SSL. To build it all you need to do is pass the --enable-ssl option to the configure script. Of course you need to have <a href="https://openssl.org">openssl</a> installed (but have you seen a system that doesn't have it?)</p>
            <p>Then I followed the documents to the letter to create the certificates and add the required configuration directives to the httpd.conf file. Yet the server simply refuses to accept connections on 443. Googling is no use. All the results are stories about alice and bob and money transfers from one bank account to another.</p>
            <p>So it was time to stop googling and to think. Thinking is great, everyone should try it sometime. While thinking I realised that you need to add the</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Apache2 SSL

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>So Google is starting an <a href="https://www.google.com/talk/">instant messaging service</a>. The really wonderfull news is that it is based on <a href="https://www.jabber.org">Jabber</a>. Three cheers for open source and four for jabber.</p>
            <p>While still on google, I finally signed up for a gmail account, primarily because it's needed to use google talk. To my great suprise I found that my nick raditha was not available. Raditha is not a very common name.</p>
            <p>And yesterday I googled for a PHP maildir implementation, to my great suprise I found that there isn't one! squirrelmail, the most popular of PHP IMAP clients does not seem to use maildir either.</p>
            <p>One of the reasons is probably that <a href="http://cy.yp.to">qmail</a> is very particular about the permissions of the maildir folders and will not deliver if the perms are not correct. That makes it difficult for PHP script running as nobody to look into a maildir folder.</p>
            <p>A couple of days ago, I started looking at python but found the tutorial rather off putting. Since then I have not looked at it again. Do you know of a good book?</p>
            <p>Each time I visit thawte I am more convinced that they are court jesters of the internet security business. Thawte offers code sigining certificates. They have an online support form with a combobox to choose the product. The comobo box only has SSL certificates in it.</p>
            <p>Looking for another bunch of jokers? look at netaddress. Many years ago I used to have a usa.net email address. Then they made it a commercial service and me being a poor old student at the time had to give it up. Now I want my old address back (mostly for sentimental reasons.)</p>
            <p>I have tried using two credit cards to place an order, heir payment system does not work. And support emails do not get answered. No wonder their shares hit rock bottom during the .com crash.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Ramblings

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Thawte proudly says everywhere on their site that 'it's a trust thing'. The question is can they be trusted?</p>
            <p>Recently I signed up with thawte's reseller program. The sign up process was protected by SSL. These guys after all are the people who issue SSL certificates, code signing certificates and all kinds of other certificates that we have not even heard about.</p>
            <p>When signing up you need to enter a username and password (duh obvious). The page is protected with SSL (equally obvious). What is not equally obvious is why they mailed me back my own password in the clear text.</p>
            <p>Can an organization that wants to you spend all your money to secure your site and then promptly mails you account information over an insecure channel be trusted?</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Trust Thawte?

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p><a href="http://fedora.redhat.com">Fedora Core</a> 4 does not install the kernel sources as it should. It' probably time to switch to Debian but in the short term I need to make <a href="http://fabrice.bellard.free.fr/qemu/">qemu</a> work but that looks like a lot of hard work.</p>
            <p>As for compiling a new version of the kernel from scratch, I have done enough of that in the to be thoroughly sick of it. That's what I ended up doing though. Merely expanding the RPM to install the source is not good enough.</p>
            <p></p>
            <div style="color:#000033">rpm -i kernel-2.6.11-1.1369_FC4.src.rpm<p></p>
                <p>rpmbuild -bp --target=noarch /usr/src/redhat/SPECS/kernel-2.6.spec</p>
            </div>
            <p></p>
            <p>There is a more recent version of the kernel but to install that would mean I cannot take any short cuts. In retrospect, I could have used the more recent version anyway because the short cut didn't work. You can see that Fredora is a Red Hat (also known as Red Face) product instead of one that is merely sponsered by Red Face from the fact that the sources were installed in some weird <em>/usr/src/redhat/BUILD/kernel-2.6.11/linux-2.6.11/</em> folder.</p>
            <p>Created a symbolic link to this wierd folder from <em>/usr/src/linux</em> and then ran the configure script in qemu. As mentioned in a previous post the buggy gcc 4 included in FC4 fails to build qemu (and many other things). So you need to use the -cc=gcc32 flag. But finally the good news is that after many hours of hard work, I have a working version of qmeu + kqemu.</p>

        </div>
    </div>


</body>

</html>

more FC4 woes

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>The last time I learned a new programming language was in 1999. That was fortran. I was working on a quantum mechanics research project at university and Fortran is the language of choise when complex numbers are involved.</p>
            <p>Unlike almost every other programming language I have seen, Fortran has a native type for complex numbers. That makes it much much faster than anything else. That's just the tip of the iceberg. If you are into numerical methods, Fortran is definitely the language of choice but fortunately or unfortunately I have forgotten most of what I once knew about fortran or Quantum.</p>
            <p>Python is the first time I am looking at a new programming language since 1999. And since I have recently gone past the 30 year mark, there is a good chance that this may be the last programming language I will ever learn. Then again I am full of suprises.</p>
            <p>The python tutorial is full of suprises too. I have read through four chapters already but I am yet to see a hello world program. Another suprise was the notation for the for loop. This looks more like shell scripting than real programming. But I am just a newbie, so ignore my comments. Wonder what will happen if I post in a message board. I have only ever asked 1-2 two questions in message boards and mailing lists but i'v responded to thousands.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Hello Python

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Two and a half years ago, when I first installed <a href="http://qmail.org">qmail</a>, I can't say I did it properly. The reason for that is the over abundance of information on installing qmail. In the end I took a shortcut and followed the <a href="http://shupp.org/toaster/">shupp's qmail toaster guide</a>, something that definitely should not have been done.</p>
            <p>This time I followed <a href="http://www.lifewithqmail.org/">life with qmail</a> and got off on the right track. After installation of qmail I spent sometime debating whether or not to install <a href="http://www.inter7.com/index.html?page=vpopmail">vpopmail</a>. The fact that vpopmail installs passwords in clear text is a bit too much to stomach but after struggling with too many .qmail files I threw in the towel and decided to go ahead with vpopmail anyway.</p>
            <p>The toaster guide and many other guides involve the creation of a separate vpopmailctl script or they run vpopmail as a standalone daemon not controlled by daemontools. Some other guides use xinetd! All that is unacceptable particularly since vpopmail will run very well under daemontools.</p>
            <p>The next step of course is to install an IMAP server, and not having had a pleasent time of it with courier I settled on <a href="http://lifewithbincimap.org/index.html/Main/LifeWithBincIMAP">BincIMAP</a>. It's shortcomings were discovered only after installation; the most annoying is that it follows a different standard for folder structre which they call IMAPdir. Unfortunately thunderbird does not like it either.</p>
            <p>After a bit of a struggle I discovered that the alternative approach described in Life with BincIMAP can be made to work but to migrate all the existing mail boxes turned out to be a bit tedius (even though they numbered less than 20). Fortunately on linux, it only takes a couple of lines of shell scripts to take the boredom out of tedius tasks.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Qmail, Vpopmail and Binc

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Since upgrading to FC4 kqemu the, accelerator could not be used (since it's not compatible with the new kernel). That effectively means Qemu cannot be used at all because with out the accelerator the emulator is too painfully slow.</p>
            <p>Unfortunately attempting to recompile again did not work out because of an error. So I downloaded the latest version (0.71) and had another go but the error simply wouldn't go away.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Qemu on FC4

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Though <a href="https://httpd.apache.org">Apache</a> 2 has been around for a long long time, I have been using apache.3 upto now, primarily because PHP can only be used with the Prefork model. But the time has come to move on or get left behind.</p>
            <p>That does not mean that this is the first time I am installing Apache 2, on the contrary I tried one of the first releases of version 2, and have been occaisionally trying out new releases just to keep in touch. This is the first time though that I will start to use it on a regular basis.</p>
            <p>Since prefork is the threading model to use, my configuration directive went something like this:</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Apache 2

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p><a href="000784.html">After overcoming several pitfalls</a> I successfully managed to boot up my Fedora Core 4 installation. Unfortunately it seems the worst is still to come.</p>
            <p><img alt="no menu" src="http://images.raditha.com/blog/images/kde-nomenu.jpg" style="float: right; clear: right; margin: 0 15px 5px 15px;">My first reaction after KDE started up was "Hey what have you done with the menu?". All the menu items have gone missing. As usuall <a href="https://www.google.com">google</a> was of no use. So I concluded that there was a bug in <a href="http://fedora.redhat.com">Fedora</a> 4 and started to update the KDE RPMS. As expected that didn't solve the problem either.</p>
            <p>Funnily enough when logged into KDE as root the menu is clearly visible so on a hunch I deleted the <em>.kde</em> folder. That unfortunately didn't solve the problem either. Then it was the turn of the <em>/var/tmp/kdecache-raditha/</em> folder. Eureka! the menu has made a belated entry onto the screen. But now having deleted the old settings I have to reconfigure my desktop all over again.</p>
            <p>Because FC4 installs a new kernel <a href="000361.html">my web cam does not work again</a>. I am giving up on it. <a href="http://fabrice.bellard.free.fr/qemu/">Qemu</a> has to be recompiled because <code>kqemu</code> can no longer be loaded. The java&nbsp;environment&nbsp;is totally screwed up because of that <em>/etc/alternatives</em> mess. Being a geek is hard.</p>

        </div>
    </div>


</body>

</html>

FC4: Hey where is my menu

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Just completed an upgrade from <a href="556.html">Fedora Core 3</a> to Fedora Core 4. The upgrade wasn't uneventfull. Things got off to a bad start when the CD device didn't get off well with the media but that was easily overcome.</p>
            <p>The first real annoyance occured when the installed refused to recognize my swap space. The swap is on a <a href="../../linux/raid.php">software raid array</a> and the installer simply refused to continue and insisted on a reboot. Which is most annoying because this is a problem that could easily be fixed by editing the fstab file. Which is just what I did when the installer started up again after a reboot.</p>
            <p>Just in case you have not done such a thing before: it's possible to bring up a shell by hitting Alt-F2 while the installer is running. Then you can temporarily mount the existing installation (if the installer has not already done so) and make any changes needed.</p>
            <p>Though the swap space was now being accepted the installer again complained about a lack of disk space on the root partition (which is absurd I have gega bytes left on other partitions). So I logged in and deleted the /usr/share/docs folder to free up about 300Mb.</p>
            <p>The <a href="http://fedora.redhat.com">Fedora</a> installer rebooted (again) and this time continued on it's merry way until the third CD was inserted when it refused to install the kdewebdev RPM which suprised me because I had the media tested.</p>
            <p>The system was rebooted yet again, and instead of booting the CD I decided to start up Fedora Core which it did but unfortunately an Xsession could not be started. So the next step was to install the RPM manually from the command line. This time an error about too many symbolic links was produced.</p>
            <p>This symbolic link issue was tracked down to the fact that I had updated the PHP documentaion for <a href="http://kdewebdev.org">Quanta</a> by symlinking to the downloaded documentation. Deleted the symlink and tried the RPM again to see that it installed with out error. Now for the upteenth reboot and this time the installer ran to completion. And after spending nearly three hours on something that should have been completed in an hour I have a running Fedora Core installation.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Fedora Core 3 to Fedora Core 4

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Finally created that <a href="../../java/jni/index.html">article</a> on mixing JNI and applets that I have been <a href="762.html">threatening to write</a>. It's the first update to the site in months and I am sure many people will find the article usefull.</p>
            <p>As I pointed out earlier, lots of pundits are of the view that you simple cannot mix native methods with applets. That is not so (as you will no doubt find). It's just a little bit complicated.</p>
            <p>The complications revolve around how the native library is to be loaded. The simplest way is to just ask the user to download the library and place it on his home folder. But that sounds clumsy so the better approach is to simply open a URL connection, read it's InputStream and get an OutputStream for saving the same library to the hard disk.</p>
            <p>This approach of course means that the jarfile has to be <a href="../../java/sandbox/index.html">signed</a>. If you are using native code the jar should be signed anyway so you need not give it a second thought.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Applets and Native methods

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>The first new version of Mega Upload since last October has just been released. It is version 0.40 of the java edition. As you can tell by the version number the <a href="../../megaupload/jsp.php">Java edition</a> has seen more neglect than the others. This new release fixes an annoying but trivial bug. More importantly it has some code improvements that will make MegaUpload easier to integrate into your code.</p>
            <p>These changes revolve around the version of <a href="../../megaupload">Mega Upload</a> that is built on the commons handler. Progress monitors for JSPUpload and Struts remain unchanged.</p>
            <p>Bug Fix:</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Megaupload-J 0.40

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>There are quite a few things about <a href="https://www.netbeans.org">netbeans</a> that makes it really attractive. Let's forget the big things like the fact that netbeans is open source etc and look at some of the small things that makes it fun to use.</p>
            <p>One of the features that I really like is that way it opens up resource bundles. When you are trying to internationalize an application having to open and switch truth several different property files is such a pain. Netbeans eliminates that pain. When you open the main resource bundle, netbeans will open the other localized versions as well and display them all as a table. What a cool and convinient way to edit the files.</p>
            <p>What else is really cool? the profiler. Once you get the hang of it, using it becomes fun.</p>
            <p>Another feature I like is the ability to load modified classes during debugging sessions.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

What's cool about netbeans

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I hate reading the perl documentation. Perhaps having the 'man' version instead of the html version contributed to my woes. But changing over to the HTML version didn't make rtfm any easier. Oh btw they forgot to add a page for sysopen</p>
            <p>Even if you had the HTML version it's organization is a poor cousin when you compare with the manuals for PHP or Java. That is not a universal truth though. Certain modules are really well documented and emminently readable. Even then you have to know in advance which module contains the functionality that you need.</p>
            <p>Then I wanted to have a break (no not a Kit Kat) but as usual forgot that perl forbids you to take breaks. Yes it's a real sweatshop. You have to use the ominous sounding <em>last</em> instead.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

More perl gripes

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>A couple of more gripes but first a bug. I have more than one project that has the same set of java files. Ideally these files should be compiled and placed in a jarfile. But we don't live in an ideal world. So instead of a single jar file I have multiple <a href="http://subversion.tigris.org">subversion</a> branches.</p>
            <p>After working on one project for a while, I closed it and switched over to another. They both have some files in common. When I did a 'find usages' search and clicked on any of the results, <a href="https://netbeans.org">netbeans</a> insists on opening the file from the earlier project. I spent more than an hour trying to figure out why my code changes had no effect. All the time I had been editing the wrong file.</p>
            <p>Now for the gripes:</p>
            <p>Is it only me or do you also find that the netbeans debugger inteprets the 'step into' as a 'continue'? This only happens when you are debugging a file it does not happen if you are debugging the main project. Apparently you can overcome this by editing the ANT build file to add a new target. If I fancied ants I wouldn't be using an IDE.</p>
            <p>Still haven't figured out why netbeans seems to completely ignore the source tree when doing a 'build' or a 'clean and build' on a web application.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

More netbeans gripes and a bug

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Many many years ago, I used to write a lot of perl code. Then I discovered PHP and thankfully I didn't have to use perl anymore. As a result I have forgotten many of the perl tricks that I learnt. Now you might say that is good news, unfortunately occaisionally I find myself occaisionally having to read/write a perl script.</p>
            <p>One of the gripes is that a language that is so good at string handling does not have a trim() function like what you find in perl. You need two lines of forgettable perl code to achieve the same result. (chomp is only good for removing the trailing newline)</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Perl Gripes

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I came across this <a href="http://www.codingforums.com/archive/index.html/t-49361.html">interesting discussion</a> about Rad Upload.</p>
            <p>My only comment is that you need to edit only <strong>two (2)</strong> lines of code and <strong>not 5 lines</strong> to get started with <a href="http://www.radinks.com/upload/">Rad Upload</a></p>
            <p>Come to think of it you need to edit only 1 line of code + 1 line of HTML.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Interesting Discussion on Rad Upload

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I have to call some native methods from inside and applet but all the pundits are saying you cannot use JNI with applets. That does not worry me at all. When we created the <a href="http://www.radinks.com/sftp/applet.php">Rad SFTP Applet</a> we heard the same thing. But that product went onto become featured on the java website.</p>
            <p>JNI is something that has been in the outer fringe of my knowledge base but I have never used it because like many I have always found more platform independent ways of doing it. Like Many I have also had trouble getting the darned thing to work first up. <a href="http://forum.java.sun.com/thread.jspa?threadID=570194&amp;tstart=45">This post in java forums sums it up.</a></p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

JNI and applets

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Had a go at at the <a href="http://profiler.netbeans.org/">Netbeans profiler</a>. The first thought that comes to mind is that it is rather strange. It will only work with the java 5 update 4 or Java 6. The java 6 early access was made available only a few days ago.</p>
            <p>Immidately upon upgrading from update 3 to update 4, Netbeans refused to start up. But that was soon put right by editing the netbeans.conf file:</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Profiling Java Apps with Netbeans

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p><a href="http://www.share-it.com">Share-it</a> as we all know is an organization that supposedly makes it easier for you to sell your software. That would be true if they knew a thing or two about customer support.</p>
            <p>One of the biggest problems with share-it is that you do not get an instant download link when you complete your order. If such a download link can be created they have done a very good job of hiding it away.</p>
            <p>The usual pattern with share-it is that you send your file to them as an attachment, and then share-it will in turn mail it to the buyer. Messy to say the least.</p>
            <p>So I asked them if there is any way at all in which an instant download link can be created. This is what they say:</p>
            <p></p>
            <div style="color:#000033; margin:15px">
                <p></p>
                <p>Thank you for your e-mail.</p>
                <p>Please send us the product as e-mail attachment if it is smaller that 5MB. We will then install it on our server, and</p>
                <p>send a download link to your customers, once we have received payment.</p>
                <p></p>
            </div>
            <p></p>
            <p>So I wrote to them again saying this is not what I asked. This is what they said:</p>
            <p></p>
            <div style="color:#000033; margin:15px">
                <p></p>
                <p>You can send us the full version of the program which can either be send out as an attachment or as an instant download</p>
                <p>link, depending on the size of the product file. We encourage you to send a zip files as customers often have</p>
                <p>difficulties receiving executable files due to firewalls or their mail clients.</p>
                <p></p>
            </div>
            <p></p>
            <p>Which is the same as the previous reply but differently worded. They are talking about sending an instant download link in an email not about allowing the user to instantly download the product. Finally after much nagging they replied that an instant download is possible but they forgot to tell me how it could be achieved.</p>
            <p>And to think these guys (along with other similar companies) charge much much higher fees than credit card processing services do.</p>
            <p></p>

        </div>
    </div>


</body>

</html>

Cut and paste and share-it

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>The first release of <a href="../../wiki/LogMine">Log Mine</a> (0.01) was added to this site a little while ago. It was decided not to rely on the services of <a href="https://sourceforge.net">Sourceforge</a>, the first release has already been <a href="000755.html">delayed</a> by more than a month thanks to them.</p>
            <p>Version 0.01 is nothing more than a proof of concept. It does not have any bells and whistles or eye candy, but it does include a working log file parser. At the moment only the common log format is supported but other formats will be added in the coming weeks.</p>
            <p>Log Mine 0.01 also includes an elementary reporting tool. It includes the seedling for the charting tool as well. The charts are primarily for comparision purposes. They are line charts that allow you to plot several items on the same set of axii.</p>
            <p><a href="https://www.raditha.com/downloads/LogMine-0.01.tar.gz">Download</a></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Log Mine 0.01 Released

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>A new version of <a href="http://www.radinks.com/upload/">Rad Upload</a> was released a few hours ago. It can now gzip compress files as they are being uploaded.</p>
            <p>A <a href="http://www.radinks.com/upload/plus/compress.php">compressed file upload</a> has many advantages the most obvious is the your bandwidth consumption drops dramatically. The end user will have to spend less time uploading whater it is that they want to share on your site.</p>
            <p>Not so obvious is the fact that compression would allow you to overcome some of the server configured limits on file upload. Of course if you are on your own server, you can simply <a href="http://www.radinks.com/upload/config.php">change the limits</a>. If not you can sneak in a larger file simply by gzipping it to reduce it's size.</p>
            <p>Other advantages include the fact that smaller uploads can be processed faster at the server and in most cases they consume less memory and of course you need less space to store the files as well.</p>
            <p>Finally Rad Upload does not rely on any external tools for the compression.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Compressed File Upload

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>It's now more than a month since I made an attempt to register <a href="../../wiki/LogMine">Log Mine</a> at sourceforge. After sitting on it for nearly a month I find that the project has been turned down. The reason? there is another project by the same name. That means I have to pick another name and try again.</p>
            <p>The existing project was registered more than two years ago. They have yet to make a release and all indications are they don't plan to make a release any time soon. the <a href="https://sf.net/">Source Forge</a> registration process gives the owners of such a project 20 days to decide if they want to give up their name. Apparently these guys want to keep it. <a href="720.html">A full month</a> has been wasted.</p>
            <p>While still on sourceforge, their new stats system seems buggier than the old one. My <a href="../../php/progress.php">MegaUpload</a> project, which gets 10 times as many pageviews and downloads as The <a href="https://sourceforge.net/projects/radnetwork">Rad Net project</a> languishes in the low 98th percentile while the latter is in the 99th percentile.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Log Mine and SF

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Often <a href="http://www.radinks.com/">we</a> are asked whether it's possible to prevent <a href="http://www.radinks.com/upload/">Rad Upload</a> from overwriting existing files on the server. The answer is indeed yes.</p>
            <p>Having said that, in HTTP file uploads the client has little say in where the file is to be saved or whether an existing file is overwritten. All that is controlled by the server side handler script. All you need to do to prevent a file being overwritten is to add a simple comparion, for example with PHP it is:</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

File Upload : Overwrite existing

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I produce here a discussion with the support staff at <a href="https://www.2checkout.com">2checkout</a></p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

jokers@2checkout

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>We just avoided being conned by a very large handheld devices manufacture. The company (that will remain nameless) tried to hoodwink us into thinking they are a small outfit. Their point of contact wrote to us using his private email address a free one and asked us to modify the <a href="http://www.radinks.net/file/file.php">Rad File Manager</a> to match their requirements.</p>
            <p>Earlier this person had downloaded <a href="http://www.radinks.com/upload/">Rad Upload</a>, when ordering he had used another email address. The customization they wanted wasn't simple but we agreed to do it for a nominal fee because we felt it would add value to our product in the long run, but the potential client insisted on an exclusive deal.</p>
            <p>We only quoted $400 for the modification - peanuts really. This guy reckoned he could tempt us with money and offered US$2000.00. What a loser, $2000 wouldn't even get close to covering our cost let alone the loss of future income. Naturally we refused. He kept nagging but we kept refusing. Then he made a slip. He sent me an email where the headers clearly indicated who he was. Tough luck for him. Either the company is trying to con us or this guy is trying to get us to do this project for peanuts and to pocket the different.</p>
            <p>I am shocked that a company that is probably worth hundreds of millions of dollars is being sneaky to say the list when paying a small business.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Handheld Con Artist

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>It's a good thing I have long hair because it's been a week of tearing at the hairs. Short ones at that.</p>
            <p>First on the 8th of june we had a spike of traffic and one of our servers crashed. This server onfortunately ran the database for the shopping cart. If that wasn't bad enough the next day there was a huge power problem at the colocation facility and all the machines in the building were constantly restarting.</p>
            <p>Unfortunatley one of their routers didn't restart, it went out and stayed out. That meant even the machines that stayed up (such as our own) were inaccessible. Now colocation facilities are supposed to be immune to power supply and network problems but as we all know 99% uptime is a big lie.</p>
            <p>Apparently the service that we use for payments have a very lousy hosting company. Frequent outages are the story of the day. The bad news is</p>
            <p>that they had another one of those on the 11th and did little about it till the 13th. They claimed it was intermittent. Well their server came up intermittently.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Server Outage.

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>The main gripe with Netbeans is that it is slow. Compiling is fast enough but that does not make up for everything else being so slow. It takes ages and ages for the code completion popup to appear and then continuing to type while it's open is painfull. Even without the code completion dialog, you feel that the text appears on the screen much slower than you type it. It may be percieved but perception matters a lot when it comes to software usability.</p>
            <p>Sure Netbeans fans will argue that it is a hell of a lot faster than eclipse. There is no denying that fact. Being faster than eclipse does not mean fast.</p>
            <p>Another gripe is with the macros, who can keep all those macros in his head? and woe betide you if you unwittingly choose to assign a macro code as a variable name.</p>
            <p>Whenever you choose to 'find usages' of a variable a dialog pops up asking if you want to search in the comments. If you have commented out large sections of your code, that means you are not a good programmer. You should learn to use a version control system instead. Having this dialog which I have never used and which I never plan to use means I have to move my mouse again. I have a sore shoulder.</p>
            <p>Last but not least I don't like this business of having a 'Main Project' sometimes you compile dozens of times pulling at your hair wondering why the code changes are not been executed only to find that it's compiling the main project instead of what you want it to compile.</p>
            <p>So what's my verdict?</p>
            <p>Netbeans is a wonderfull IDE. It just needs to sort out a few glitches.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Netbeans Gripes

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Continuing <a href="671.html">the saga of Java 5</a>: The combined netbeans/java 5 bundled download seems to work very smoothly. Is it the so called seemless integration or have the managed to fix a few bugs? God knows <a href="000532.html">the first release had more than it's fair share.</a></p>
            <p>However, trouble occured after a test project had been built. When trying to use <a href="http://proguard.sourceforge.net">proguard</a> to optimize the jar file I kept running into an error</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Unsupported version number [49.0]

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>The java 5 update 1 was released a few months ago. It was not long after our country was <a href="597.html">ravaged by the tsunami</a>. Even though the update was downloaded, I never had the chance to try it out. Days turned to weeks and weeks turned to months and update 1 became update 3. Finally I decided to download it and here are the comments.</p>
            <p>The JDK-Netbeans combo download itself if 115Mb, expanding and installing it was a breeze thanks to the gui installer, and the installer seems to be pretty quick too.</p>
            <p>My installation of previous versions were hardly pleasent <a href="548.html">all manner of errors</a> come up when attempting to compile/run applications with it. From the surface they seem to have gone away with this new version.</p>
            <p>I always prefer to keep class files and source files separate and in fact the best place for class files is the /dev/shm folder. When you place your classes in this folder what you are really doing is to keep them in memory without writing them to disk. That makes the compilation go that much faster (provided of course that you have lots of memory).</p>
            <p>It isn't exactly obvious how you can change the output directory in Netbeans but I finally managed to track it down to the <em>build.dir</em> property in the /nbproject/project.properties file. Now to put this to the real test.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Java 5 update 3

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>The other day, I wrote about having trouble <a href="728.html">importing a thawte certificate</a> back into the keystore. No matter what the certificate refused to add it self to the store - or if it was added the jarsigner refused to recognize it. The only solution seems to be to get a reissue.</p>
            <p>Thse are the exact steps:</p>
            <p>First create a new keystore. Creating a new key in an empty keystore does the trick.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Renew Code Signing certificate

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Just renewed <a href="http://www.radinks.com">our</a> code signing certificate. Having heap of trouble importing it back into the keystore.</p>
            <p>First try:</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Keytool, CSR and certificate replies

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Transparent proxy or interception proxying as it is sometimes known is easy enough to setup. All you need is a simple IPtables rule:</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Transparent proxying for the localhost

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>It is said that a <a href="http://fabrice.bellard.free.fr/qemu/">qemu</a> guest operating system can talk to it's host operating system through the network with the help something called <a href="http://vtun.sourceforge.net/tun/faq.html">tun/tap</a>, something which I am ashamed to say I have never heard of before.</p>
            <p>Configuring network cards, ppp or routing tables has never been my strong point, so TUN/TAP isn't something that I approached with eagerness. But having <a href="721.html">successfuly setup qemu</a>, networking is sorely needed if it is to be made full use of. It's just too bad that there isn't anything really usefull on <a href="https://www.google.com">google</a> about this topic.</p>
            <p>After struggling unsuccessfully to asign an 192.168.x.x IP to the guest operating system (the NIC on the host system has such an IP), I decided to rever to the default 172.20.x.x ip (which is in fact asigned by DHCP). The original qemu-ifup script was also reverted to. It assigned 172.20.0.1 to the tun0 interface on the host.</p>
            <p>I can open a socket connection in 5 different programming languages but ask me to set up a NAT and I would struggle. Having said that I still needed to enable NAT so that the guest system could access the outside world. Fortunately, I can type in :</p>
            <p>iptables -t nat -A POSTROUTING -o eth0 -j MASQUERADE</p>
            <p>and</p>
            <p>echo 1 &gt; /proc/sys/net/ipv4/ip_forward</p>
            <p>by rote. There is still another issue to be addressed; I have a webserver on the host system with about a dozen IP virtual hosted sites. Since all the IPs are private network IPs they do not have DNS entries, that meant the hosts file on the guest system had to be edited. That's only a part of the solution becuase it's not practical to create a dozen virtual NICs every host entries, all host entries point to the same IP. That calls for a another set of iptable rules.</p>
            <p>First I try adding rules to the PREROUTING chain, then after tearing my hair tried the same with the OUTPUT chain still without success, finally I jumped to the wrong conclusion that tun0 is not too concerned about obeying iptables rules. That was proved wrong by the simple rule</p>
            <p>iptables -A INPUT -i tun0 -j DROP</p>
            <p>After struggling with it a little longer, I finally realised that had I chosen to set up a few virtual nics I would have finished it in less than 5 minutes. This is of course the down side of being a geek.</p>
            <p></p>

        </div>
    </div>


</body>

</html>

Qemu and networking

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Earlier this month, I wrote about the need to <a href="000715.html">create a different web log analyzer</a>. I have started on it. A new project has been created at <a href="http://www.sourceforg.net">sourceforge.net</a> and the first release will be added in a couple of days time. Interestingly that would coincide with the new sourceforge stats system coming back online.</p>
            <p>One of the main feel good factors about open source projects is knowing how many people use your product, with the sourceforget stats being off line for more than four months, this simple pleasure has been denied to thousands of developers. Now the sourceforge stats and back and they definitely look worth the wait. But I digress.</p>
            <p><a href="../../wiki/LogMine">Log Mine</a>, as the new project is called, is just getting started. So the first release will be nothing more than a proof of concept. But it does have a working log file parser. The parsed records will be inserted into a relational database, which make it possible to run all kinds of queries against it and generate all manner of reports and charts.</p>
            <p>In fact a very very basic charting tool is also included. Unlike traditional charting tools of many a web log analyzer, Log Mine can plot several different 'resources' on the same axis. This allows you to compare (as an example) the requests for site entry page vs checkout page or sign up page etc.</p>
            <p>As usuall with all my projects (including those at <a href="http://www.radinks.com/">Rad Inks</a>, the download is very small. Nothing I hate more than bloatware.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Log Mine

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Just downloaded 0.7.0 of <a href="http://fabrice.bellard.free.fr/qemu/">qemu</a>. Looks like there have been two releases since the last time I tried it out. Improvements include a step towards <a href="https://www.apple.com/macosx/">Mac OS X</a> support.</p>
            <p>Last time around I didn't have much success with it partly due to a lack of patience on my part. The documentation being very cryptic didn't help either. This time around, I found that there is a running commentary on qemu on varius sites and more importantly <a href="http://www.advantageplayer.com/blackjack/forums/bj-tech/webbbs.cgi?noframes;read=2043">answers!</a></p>
            <p>So unlike the last time around I had more success in installing the guest operating system. This is the command I used at first:</p>
            <p>qemu -boot d -cdrom /dev/hdc -hda hd.img</p>
            <p>Unfortunately it turned out to be painfully slow. So I tried copying the entire CD into <a href="000527.html">memory</a></p>
            <p>dd if=/dev/hdc of=/dev/shm/boot.img</p>
            <p>Of course you need to have a lot of memory if you take this approach, it only lead to a small increase in speed, but it was enough. Once the installation was completed, this is how I boot the system:</p>
            <p>qemu -boot c -hda hd.img</p>
            <p>Now I feel like ordering a copy of OS X Tiger and trying that out as the guest system.</p>
            <p></p>

        </div>
    </div>


</body>

</html>

Qemu

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I have complained many times about the <a href="703.html">lack of a site filter for Adwords</a>. As a result if you choose to advertise on the <a href="http://adwords.google.com">Adwords</a> content network you have <a href="679.html">no control of where the ads were being displayed</a>. Or at least that is until last week you had no control. Now at long last <a href="https://www.google.com">google</a> has finally added the ability set campaign wide 'site filters'</p>
            <p>At <a href="http://www.radinks.com/">Rad Inks</a> we immidiately made use of it to see our daily spending go down sharply. In spite of the decrease in advertising cost, we have seen a marginal increase in sales figures, so obvioulsy site filtering is a good thing.</p>
            <p>The fate of many sites on the <a href="https://www.google.com/adsense/">Adsense</a> network remains to be seen, more and more sites will surely get into one filter or another and will be reduced to displaying 'remanant campaigns'</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Adwords Site Filtering at last

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>No one would want to build yet another log analyser. There are dozens of them out there. Things are not as bad in the in the field of Content (Mis)Management Systems where everyone and his dog seems to have a product listed at freshmeat but there are too many log parsers out there.</p>
            <p>So why do I even think of doing my own web log analyser? I am not. I am thinking of building a data mining system for log files. Sure you will be able to use to figure out how many times your baby on the lawn pictures were seen, as long as you don't mind waiting for dozen queries to be executed.</p>
            <p>The idea I have in mind is all about ratios. Webdruid and other try to show you graphs supposedly of the path that users took through your site. As analog's authors have taken great pains to explain such reports are unreliable at best.</p>
            <p>What is reliable and what works is ratios. Let's take an example; the success of your ecommerce site can be meatured in the ratio of homepage views/product views/add to cart/checkout/finalize or some such order.</p>
            <p>Trouble is, with many web log analyzers you need the help of a calculater or spreadsheet to get to these ratios. If you want to plot ratios against time or some other factor it takes a lot of work, and how do you compensate for seasonal slumps/spurts?</p>
            <p>The answer lies in using a database. Entering the webserver log files into a datbase directly is woefully inefficient. For a site that delivers only a few thousand pages a day you might need to put the database in it's own server, if you took this approach. For simple logging there is nothing like a text file - which is just what your good old apache log file.</p>
            <p>Traditional web log analysers are all about telling you how many pages a day were delivered and which pages were most seen etc etc. For such tasks you don't need a database. Suppose on the other hand that you want to generate reports with what ever data, what ever period that takes your fancy and not be contrained by the rigid daily/weekly/monthly structure of the traditional log analysers. Then you need a db.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

To build another log analyser

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>For the third day running, the topic is log analysis tools. Earlier I wrote that <a href="000710.html">webalizer/webdruid were unsuitable</a> for my requirement because they refused to process my logfiles properly. It turns out that the real issue is not the logformat but the sequence.</p>
            <p>In whodunnits the best suspect usually comes out smelling of roses in the end. When a web log analyzer does not produce the expected result the fingure is usually pointed at the logformat and jump to the conclusion that it is not being detected properly. In some tools the order of the log entries is more important.</p>
            <p>For example if the tools first sees a record for May 1 and then subsequently sees a record for for April 30, it will choke. Fortunately there is a solution.</p>
            <p><a href="http://awstats.sourceforge.net/">Awstats</a> comes to the rescue of <a href="http://www.mrunix.net/webalizer/">webalizer</a> and <a href="http://www.webdruid.org/">Web Druid</a>. It (awstats) contains a script named <em>logresolvemerge.pl</em> that will sort the logs into the proper order. The output produced can be safely passed into either webdruid or webalizer (and even awstats) itselfs and the files will be properly processed.</p>
            <p>Another drawback of webalizer and it's fork webdruid is that you cannot set any filters for status code. For example in <a href="http://www.analog.cx/">analog</a>, you can produce a report for all redirection requests (usefull if you have restructured your site). The web* combo does not produce such a result and does not even display the redirected requests properly in the <em>all urls</em> page. To overcome this, I made a slight mod to the webdruid.c file. Let me know if you want this updated version.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Awstats rescue Webdruid and Webalizer

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Yesterday, I took a look at <a href="000710.html">three big names in web log analysis</a>; <a href="http://www.analog.cx/">analog</a>, <a href="http://awstats.sourceforge.net/">awstats</a> and <a href="http://www.mrunix.net/webalizer/">webalizer</a>. Today is the turn of some less well known products that are listed on <a href="http://www.freshmeat.net">freshmeat</a>.</p>
            <p>There are hundreds, so only the products that do not have the appearence of abandonware were anlysed. That is only products that have released a new build during the past few months were looked at.</p>
            <p>The first such was <a href="http://www.hping.org/visitors">Visitors</a>, a nice little 70kb download. Installation was a piece of cake (just type make). Running it though is not quite the same because <em>Visitors</em> does not seem to have a configuration file. All the settings have to be passed in as command line parameters. Secondly it does not allow you to change the logfile format, but since it recognized my own files I shouldn't be complaining.</p>
            <p><em>Visitors</em> has two features to help you figure out what 'trails' a visitor took through your site. As the author of analog takes great pains to explain such reports are unreliable and downright misleading. The second feature (related to the first), is the ability to produce a graph of the visitor's path. It requires the <a href="https://www.graphviz.org/">graphviz</a> package to be installed.</p>
            <p>Next I tried something called logmine. Words fail me.</p>
            <p>Then I tried a program called <a href="http://www.nettrackerlite.com/download/">Nettracker</a>, it claims to be an award winner. Who ever gave them the award didn't try their installer.</p>
            <p>The next victim was <a href="http://www.webdruid.org">Web Druid</a> a webalizer fork, and that sounds promising. Webdruid can also interact with graphviz to show you which paths the user took through your site. I didn't bother to find out if this feature was infact available in webalizer.</p>
            <p>Looks like another study, this time a comparision of webalizer and webdruid is in order, but first they need to be persuaded to accept my log files.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

The less well known log analysers

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p><a href="http://www.analog.cx/">Analog</a> has been my web log analyzer for as long as I can remember. There are several reasons, first and foremost being that it was the best around many many years ago, and once you find a good tool you stick with it. I have stuck with it for a long time.</p>
            <p>That does not mean I have been faithful, I have looked at <a href="http://www.mrunix.net/webalizer/">webalizer</a> and <a href="http://awstats.sourceforge.net/">awstats</a> on a couple of occasions. Just a little while ago, I tried webalizer using the version that was on my Fedora Core 3 CD. Unfortunately webalizer does not allow you to specify the log file format. It wants you the change the log format in your webserver configuration. This is a big nono if you have CD loads of logs in a particular format. So naturally it's time for <em>rpm -e webalizer-2.01_10-25.i386</em></p>
            <p>The last version of Awstats that was installed on this computer was 5.9.1 which was painfully slow. That is only to be expected it is written in perl after all while webalizer and analog are created with C. Slowness wasnt' the only problem, it took considerable effort to persuade it to accept my log format, and awstats wasn't impressed by the fact that the log files were gzipped.</p>
            <p>Though under normal circumstances, I download development versions instead of stable versions, now I decided to get 6.4 version of Awstats instead of the 6.5 branch. That is because the 6.5 branch does not yet include an RPM, last time around installing from source was very painful affair and that mistake wasn't going to be repeated this time around.</p>
            <p>Even installing from RPM is not as simple as it sounds, there are three additional steps to follow, so let's digress for a moment. I am occaisionally look out for alternatives for analog because though it is fast and accurate it has one shortcoming. It will not give you monthly stats on a page by page basis. The manual is rather cryptic (the other manual are really no better) and updates are infrequent.</p>
            <p>Configuring awstats is quite easy unless you have a custom log format, then it becomes a bit more challenging but the same can be said about analog. The next challenge is to get the analyzer to read your files in chronological order. The logresolvemerge.pl tools is your friend.</p>
            <p>Now that you have put up with this tirade, let me give you something usefull. here is a sample from my log file:</p>
            <p>66.196.90.194 - - [16/Mar/2005:16:04:11 -0800] "GET /robots.txt HTTP/1.0" 404 272 "-" "Mozilla/5.0 (compatible; Yahoo! Slurp; http://help.yahoo.com/help/us/ysearch/slurp)" raditha.com</p>
            <p>This is pretty much a standard log format except for the fact that the virtual hostname is appended to the end of the record. Now here is the analog config directive:</p>
            <p>LOGFORMAT (%S %j %u [%d/%M/%Y:%h:%n:%j] "%j %r %j" %c %b "%f" "%B" %jn)</p>
            <p>And here is the webalizer directive:</p>
            <p>Does the line look blank? yes it is blank, as I already mentioned above you cannot change the format, now here is the awstats version:</p>
            <p>LOGFORMAT (%S %j %u [%d/%M/%Y:%h:%n:%j] "%j %r %j" %c %b "%f" "%B" %jn)</p>
            <p>Finally please ignore the fact that this entry is long, very long, probably the longest entry for this blog. Awstats takes a very very very long time to run.</p>

        </div>
    </div>


</body>

</html>

Analog Webalizer and Awstats

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I have not dabbled in any new technology for a couple of months. I have not even been close to the edge and readers of this blog and visitors to this website do not appreciate that. It is reflected in the site statis for April, which shows a drop in april after an almost continues rise for the last few months.</p>
            <p>So in order to put things right, I decided to download and compile php 5.04 now you can hardly call that as anything close to the blunt edge, let alone the cutting edge, but it has it's uses. When you dabble in new technology on an almost daily basis trying out new things come naturally, after been away from it all for about two months it looks somewhat less appealing.</p>
            <p>Doing a good old fashioned ./configure, make, make install is just the right thing to get into the groove. Even had to solve that <a href="000609.html">*** glibc detected *** double free or corruption</a> all over again, which was fun because I already knew the answer.</p>
            <p>You know what I am talking about, man is a creature of habit, there are certain that that we do the same way everyday. We like routine.</p>
            <p>So what upset my routine? well, lots of things including the fact that April is the big holiday season in Sri Lanka, the very hot weather that plagued us for a couple of months and now the rains!</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

The blunt edge

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Just made a <a href="http://www.radinks.com/upload/changes.php">new release</a> of <a href="http://www.radinks.com/upload/">Rad Upload</a>.</p>
            <p>The other day, I wrote about changes to Rad FTP in order to make it compatible with some FTP servers on the windows platform. Well, Rad Upload has it's built in FTP client. As you might guess it is based on some of the classes from Rad FTP. So it stands to reason that a new release of our file uploader has to hit the shelves.</p>
            <p>Improvements to the FTP functionality is not the only change though. We have also made improvements to the <a href="http://www.radinks.com/upload/plus/image.php">image scaling</a> capabilities of the applet.</p>
            <p>Rad Upload can be configured to scale images larger than a certain height or width before they are uploaded. From version 2.00 (when this feature was introduced) to version 2.02, you had to enter both a height and width. Some clients however told us that there are certain instances where it is inconvinient to specify both dimensions. So we listened.</p>
            <p>The new version of Rad Upload Plus will allow you to scale images by entering a max for only one of the two dimensions. The other can simply be left undefined, or set to -1.</p>
            <p>We were hoping to jump straight from version 2.02 to 2.10 (which will include gzip compression), but we thought these changes significant enough to warrant another release in between.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Rad Upload

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>The <a href="https://adwords.google.com">adwords</a> home page proudly says: "Pick the exact sites where your AdWords ads appear". This is actually a misnomer. It is <a href="https://www.google.com">google</a>'s way of avoiding a problem with adwords.</p>
            <p>As you know you can choose to display ads on the so called content-network. The content network is made up by websites that have signed up with the <a href="https://www.google.com/adsense/">adsense</a> program. The name (content-network) does not fit very well. Some of the sites do not have any content at all. With other sites it's blatantly obvious that many pages were designed not to be usefull to anyone but simply to hold a couple of banners.</p>
            <p>What is suprising is that the adwords program does not allow advertisers to filter out such websites. Google does not seem too inclined to do filter them out themselves. As a result, those who choose to advertise on the content network will have their ads displayed on all manner of sites, <a href="679.html">even the ones they do not want to advertise on</a>.</p>
            <p>Google recognizes that there is a quality issue, that is why they charge a lower rate for the click throughs recieved from content network sites, but often that is not enough becuase some low quality sites still manage to produce a large number of click throughs. Besides which the lower payouts effect everyone not just the bad sites.</p>
            <p>At <a href="http://www.radinks.com/">Rad Inks</a> we studied a random sample of 1000 clickthroughs and found that 542 of them originated from the same website - a site that we do not want our ads to appear on. The only option is not to use the content network at all, but then we do not get the desired amounts of traffic.</p>
            <p>Now to get back to the "Pick the exact sites where your AdWords ads appear" statement. This just means that you will soon have the option of using CPM advertising instead of CPC advertising. Since CPM ads will run in comptetition with CPC advertising, chances are this will not lower your advertising budget.</p>
            <p>Unless google allows some really low (I am talking about </p>
            <p></p>

        </div>
    </div>


</body>

</html>

CPM adwords

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p><a href="https://www.hotscripts.com">Hotscripts</a>, <a href="http://www.download.com">download.com</a> and <a href="https://www.sourceforge.net">Sourceforge</a> seem to be heading in opposite directions.</p>
            <p>Until recently hotscripts did not discriminate between commercial and free products. According to their latest updates, unless you cough up the fees for a premium listing ($38/quarter last time I checked), commercial products are going to be relegated to the bottom of the heap.</p>
            <p>CNET/Download.com is heading in the other direction. Though popularly thought of as a shareware site, only commercial products were listed in it for the simple reason that you have to cough up a minimum of $9.00 per month to be listed in their directory. That however has changed.</p>
            <p>Now you can list products (be they commercial of free) free of charge on CNET. Paid listings get better placement at all that but at least now open source vendors will be able to get their products listed on it. As chance would have it this change has just come about after I renewed almost all our listings on their directory. Some of the listings do not generate much in the way of traffic and had the free option been available a month ago, I would have jumped at it.</p>
            <p>And what of sourceforge.net? it's going now where. Their statistics pages have not been updated for more than three months, yet inspite of that they somehow manage to pick a project of the month. The only way they can possibly do that is by drawing lots.</p>
            <p></p>

        </div>
    </div>


</body>

</html>

Hotscripts, Cnet and Sourceforge

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>A new version of the <a href="http://www.radinks.net/ftp/applet/">FTP applet</a> was released on the 27th. <a href="http://www.radinks.net/ftp/changes.php">What's new in this version?</a> not much except for a fixing up a couple of compatibility issues.</p>
            <p>FTP servers running on window the copy cat system can sometimes be configured to return a unix style directory listing. (why?). In most cases the Applet handled this situation well, however there were still a few cases where it wasn't handled properly (would you blame the applet?). The motivation for this new release was primirily to solve this issue.</p>
            <p>Secondly we noticed that the applet did not work as expected on titan ftp servers so we fixed it too at the same time. Titan is in fact both an SFTP and FTP server (which is unusual). And guess what <a href="http://www.radinks.com/sftp/">Rad SFTP</a> does not work too well with it's SFTP servers. yikes.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

FTP Applet

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>My trusty old Siemens S55 has an item called 'GPRS Info' it says a total of 1016.87 MB has been transferred. That's 1GB, if you consider the fact that the sim has spent quite a bit of time inside other phones such as the 6600, the 6630 or the P910i discussed here the total would be much much higher.</p>
            <p>I have no doubt that this is the highest usage by anyone in Sri Lanka - in fact I intend to find out.</p>
            <p>(since the previous post was rather lenghty, this post was kept short, so that nature's balance is preserved)</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

1GB of GPRS

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Companies without a US office are at a huge disadvantage when it comes to credit card processing. If you happen to be a Sri Lankan company, accepting credit card payments is like swimming upstream with both hands tied behind your back.</p>
            <p>When choosing a payment gateway options are really limited for non us companies because most gateways are available only if you have a US bank account - for which you need to have a US presence. Something that is too costly for most small businesses.</p>
            <p>At first glances it would seem that there are plenty of choices, that is until you actually try to sign up with a gateway. The sign up forms do not even have a field to enter the country.</p>
            <p>There are several others namely, <a href="https://www.2checkout.com">2Checkout</a>, <a href="http://www.paysat.com">Paysat</a>, <a href="https://www.ccavenue.com">CCAvenue</a> and <a href="http://www.ikobo.com">Ikombo</a> that will accept international merchants. There is always a catch. CCAvenue will only accept vendors selling tangible goods. Paysat and IKobo do not accept all major cards.</p>
            <p>Here in Sri Lanka, where the local banks which pay 5% interest to depositers and have prime lending rates in excess of 10% boast of payment gateways on their website. But you cannot sign up online and when you call to find out, no one seems to know anything about it.</p>
            <p>Faced with all these problems <a href="http://www.radinks.com">we</a> have been using 2checkout for the past two years. Unfortunately their service standards have dropped over the last few weeks like a ton of bricks. Curiously if you search the web you will find lots of people saying it has always been that way.</p>
            <p><a href="https://www.paypal.com">Paypal</a> is not an option. Though they say "send money to anyone with an email address" that is a lie , their service is available in only 40 countries and our country is not in that list - which is perhaps fortunate jugding by some of what has been said about paypal.</p>
            <p>In order to find a more suitable solution I had a chat with a financial consultant who has lived and worked in several countries. He pointed me to several Sri Lankan companies who sell online and find out how they do it.</p>
            <p>Unfortunately these organizations primarily target Sri Lankan consumers and sell tangibles. What use is a consultant that cannot tell the difference between a tangible and intangible product? I didn't tell this consultant that I processed my first credit card transaction in 1997 long before most of the other Sri Lankan companies even hear d of the internet.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Payment Gateway Woes

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>Time for another complaint against <a href="678.html">the nokia 6630</a>. To be fair it may not be a problem with the device itself but a problem with the memory stick that accompanied it.</p>
            <p>When you attempt to access either the still or video camera you get a nasty message that just says 'System Error'. This happened at a time when I was out of town and without ready access to the net. The usuall remedies of rebooting, taking out the battery+sim to reboot again does not work.</p>
            <p>Damned if I will reformat the whole thing again like I did the last time the phone caused problems - there are a couple of photos that I wish to salvage. The next attempt was to remove the MMC and reboot the phone. Viola! with out the memory, the camera (both the still and video) can be readily accessed - even though the phone compains about the missing memory, it will automatically switch to the built in phone memory and continue to function without failing.</p>
            <p>Now to plug in the MMC again and try to access the camera, as you may have already guessed you get nothing more than that boring old 'System Error' again.</p>
            <p>This probably means I am going to loose the photos - this fear was confirmed when the device was plugged into a computer with a USB cable - everything in the device is inaccessible, the darned thing has to be formatted.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

6630 System Error

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            As the subversion documentation says, a good FAQ is a list of questions people might actually ask and not a list of questions some none geek marketeer dreamt up. These are some questions and answers based on my own experience.

            <dl>
                <dt>Oops, I entered the wrong text when commiting my changes how do I change it?
                </dt>
                <dd>If you don't remember the revision number for your commit, type <em>svn info</em> to find out then. look in the db/revprops/ folder in your subversion repository. Assuming this is at /var/subversion and your revision number is 122 the file /var/subversion/db/revprops/122 will contain the message that you typed in. You can open this file in your editor and change the message but don't edit anything else.
                </dd>

                <dt>Can you do a stat against a specific revision?</dt>
                <dd>the <em>svn stat</em> command certainly does not accept the <em>-r</em> parameter through witch you can generally tell subverison which revision you want to compare against. The simplest solution would be check out the revision of interest into a clean directory and diff against that.</dd>

                <dt>How Do I revert to a very old revision?</dt>
                <dd>
                    <p>The <em>svn revert</em> command, like the <em>svn stat</em> command does not accept the -r parameter. Fortunately the <em>svn update</em> command does. So instead of reverting you can actually update to an older version (now doesn't that sound like a contradiction?). But it's probably going to give you problems when you try a commit.
                    </p>
                    <p>Another approach is to remove your current working copy and start off with a fresh checkout - of an older revision.
                    </p>
                </dd>

                <dt>I try to make a branch and I get this error:
                    svn: Commit failed (details follow):
                    svn: Invalid change ordering: new node revision ID without delete
                </dt>
                <dd>I usually see this error when I try to make a new branch/tag using the working copy and I have some locally modified files in the working copy. Subversion should perhaps give a different message that indicates the root cause (the local modifications). Once you commit your changes you can easily use the <em>svn cp</em> command to create your branch.
                </dd>
            </dl>


        </div>
    </div>


</body>

</html>

Subversion FAQ

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I was just checking up on who links here when I realised that message boards and newsletters are getting more of a wild west character each day. Or rather it's a case of the blind leading the blind.</p>
            <p>It is for this reason that I have not posted anything on the PHP mailing lists for a long time. The PHP-General mailing list is populated by a bunch of idiots who believe it to be the 'General' mailing list. Old timers who have attempted to keep the posts on topic have been sharply rebuked by people who believe it's our job to help people who are too lazy to google or RTFM. As a result most of the replies are now made by these very people and not much needs to be said about the quality of the replies.</p>
            <p>As for the message boards, I found a discussion on <a href="692.html">gilbc detected *** double free or corruption: 0x050ca9d0 ***</a> The second post on that thread points to my <a href="000609.html">post on jan 11</a> after that it gets pretty wild. Some even suggesting that these errors are caused by faulty hardware.</p>
            <p>If your RAM is faulty or your <a href="52.html">CPU is toast</a>, your machine simply will not boot. Period. If by some chance only a few transistors in the RAM are busted and your machine manages to boot up, the moment that area of RAM is written to your machine will go down like a ton of bricks. Those unfortunates on windows systems will see the notorius blue screen of death. The only hardware failures that ordinary desktop PC can handle well is a hard disk with bad sectors.</p>
            <p>As <a href="http://www.jasonn.com/node/107">Jason - a blogger who has linked here</a> as pointed out, there are all sorts of people throwing darts in the darks pretending to know what they are talking about, more often than not it's just a matter of bluff. Jason was referring in particular to my post on <a href="000661.html">SSHD</a>.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

The Message Board Wild West

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>I have in my hands an Ericsson 910i and I am wondering what all the fuss is about. This is a phone that tries too hard to be too many things at the same time. This expensive phone cannot hold a candle to the <a href="678.html">6630</a> - the other phone I am toying with at the moment. The phone I use on a regular basis and the only one I own is an old siemens S55, and that's far superior to the 910i</p>
            <p>For example the 910i has a qwerty keyboard but the keys are so tiny it's nearly impossible to make use of it while holding the phone with one hand. I recall seeing other phones that allow you to bring up a keyboard on the display and tap at it with the styles, perhaps this phone has that feature but I am yet to find it.</p>
            <p>How about the camera? the camera is of appalling quality, it's nearly impossible to use it indoors the pictures are of appaling quality, the resolution is only 640x480.</p>
            <p>It does have an MP3 player but without the headset it sounds like a scratched up old record. It does have a video recorder and player but it looks grainy even on the phone (haven't tried playing the files on the PC yet).</p>
            <p>What about J2ME? I haven't even figured out which version of MIDP is installed on it yet (the manual didn't say and I haven't checked up the site yet - Sony-Ericsson like Siemens does a very good job making information hard to dig up).</p>
            <p>Does the phone have any positives? yes. It has huge display and it has a built in chess program which I quite liked.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>

Ericsson 910i

<html>

<head></head>

<body>
    <div class="item">
        <div class="item-header">



        </div>

        <div class="item-body">

            <p></p>
            <p>A few days ago, I wrote about the <a href="http://www.radinks.net/file/file.php">Rad File Manager</a>. At the time it was a Java File Manager. That has not changed, Janus had two faces, now Rad File Manager has two backsides!</p>
            <p>Well jokes apart; Rad File Manager as a newly added PHP web application. As I mentioned in my previous post, this is a software that is made up of a web application which runs on your HTTP server and a client that runs in your browser. They work together to provide the functionality that you might find in an SFTP or FTP client/server combination.</p>
            <p>For the previous version (1.00) the server side components were in java (Servlets and JSP) for the new version (1.10) we have made recreated the same <a href="http://www.radinks.net/file/php-manager.php">web application in PHP</a>.</p>
            <p>So why have two 'editions'? Well web developers fond of java will want a java version and a php scripters will want a PHP version. At <a href="http://www.radinks.net/">Rad Inks</a> we try to keep our design as simple as possible - and the design of Rad File Manager is simplicity itself. That's what makes it so easy to integrate it into existing web applications.</p>
            <p></p>
            <p></p>

        </div>
    </div>


</body>

</html>