May 16, 2006, 7:11 a.m.
I was trying to run a .NET command line application on Windows Server 2003. It worked on my Windows XP development machine, but when I ran it on the server, I got a file not found exception. Huh, I must be missing a component. The exception didn't say which file wasn't found, so I installed a couple packages that weren't on the server. Nope, still get a file not found exception. Ok, time to figure out which file is causing the problem. <a href="http://www.sysinternals.com/Utilities/Filemon.html">Filemon</a> is great for this. You can filter on the name of your executable and see every call to the file system. I then look at all the NOT FOUND files. I could probably create a better filter, as a lot of the files that are initially NOT FOUND usually get resolved (it's pretty educational to see where the system looks). This time, I noticed a file, call it FOO~1.DLL, that was NOT FOUND. It looked like it never got opened with a result of SUCCESS, either. Hmmm, what about that abbreviated 8.3 filename? Turns out you can <a href="http://www.sysinternals.com/Utilities/Filemon.html">disable</a> <a href="http://support.microsoft.com/kb/121007/">the</a> <a href="http://www.windowsdevcenter.com/pub/a/windows/2005/02/08/NTFS_Hacks.html">creation</a> of abbreviated filenames in Windows Server. I uninstalled the original app I was trying to run, enabled 8.3 filenames, rebooted the server, reinstalled the app, disabled 8.3 filenames and rebooted the server, and I was able to run it. My question is, why does my .NET 1.1 app search for the dependency with a short filename?
May 10, 2006, 11:21 a.m.
Usually, when I build a Linux dev box, I turn off the firewall. This time, I left it on with the default settings. I installed <a href="http://www.jboss.org">JBoss</a> and couldn't figure out why I couldn't browse to any of the ports (dumb dumb dumb dumb dumb!). Eventually I remembered I had a firewall running. I hadn't done anything with Linux firewalls for a while, but iptables is pretty easy to use. Of course, it's not like I had to do anything incredibly complex either: <blockquote> sudo iptables -I INPUT -p tcp --destination-port 8000:8999 -j ACCEPT</blockquote> This just accepts incoming connections on 8000-8999.
May 5, 2006, 2:43 p.m.
As previously noted, I installed FC5 in a VMWare image. I should have configured it to be a server, but decided to see how much running X affected performance. Well, I noticed a little lag in my xterm window and ran top. The image was running at a load of about 1, with X being the culprit. I think (seriously) it might have been spent those cycles drawing the screensaver in my VMWare window, so I decided to just not run X. I've spent the last 18 months doing mostly Windows work, with some simple Linux tasks here and there. Like Stan and Kyle, I learned something today: these days, the measure of success is not how long it takes you to remember how to do a task, or read the man pages, it's how quickly can you craft a Google query to find the right page. I'd forgotten that runlevel 5 starts X, so you can either run at runlevel 3 or modify /etc/inittab (which is what I ended up doing). It took me three different queries to get to "remove gdm." I think the third entry in that result list talks about exactly what I wanted to do. Also, it looks like Fedora now uses X to display boot info? Interesting. <strong>Update:</strong> looks like Yum is part of the default packages installed (for whatever configuration I picked, think it was development). Nice, I like yum. Now, if I can get it to work...it just hangs when I run it. Wonder if it's something to do with VMWare? Or the fact that I haven't installed VMWare Tools yet (there are conflicts with VMT and FC5 that I don't feel like fixing right now). <strong>Another update:</strong> if you try to install the J2SDK 1.4.12 (and probably other versions), it will fail with a message like this: <blockquote>Unpacking... tail: cannot open `+511' for reading: No such file or directory Checksumming...</blockquote> After doing some Googling, I found the <a href="http://forum.java.sun.com/thread.jspa?threadID=712526&tstart=0">appropriate thread in the Java forums</a> and the <a href="http://fedoraproject.org/wiki/Docs/Beats">relevant Fedora docs</a>, which say this: <blockquote>The <tt>coreutils</tt> package now follows the POSIX standard version 200112. This change in behavior might affect scripts and command arguments that were previously deprecated. For example, if you have a newer system but are running software that assumes an older version of POSIX and uses <tt>sort +1</tt> or <tt>tail +10</tt>, you can work around any compatibility problems by setting <tt>_POSIX2_VERSION=199209</tt> in your environment. Refer to the section on standards in the <tt>coreutils</tt> info manual for more information on this. You can run the following command to read this information. <pre>info coreutils Standards</pre> </blockquote>
May 4, 2006, 2:56 p.m.
A lot of the new <a href="http://www.thrall.net/maps/mbta.html">web</a> <a href="http://www.zillow.com">applications</a> that have come out in the last few years have extracted meaningful information from multiple sources and presented it in a way that is very useful. I think we will only see more of these applications. One problem that needs to be solved again and again is how to infer meaning from unstructured data. How do we get the information we want to display from a site? I'm starting to see more RSS feed generators. <a href="http://www.feed43.com">Feed 43</a> was the first one I saw. It takes the approach of downloading the source for a page, then allowing you to create filters for what you want to scrape. This is probably the best approach for now. <a href="http://www.ponyfish.com">Ponyfish</a> is another site you can use to generate a RSS feed. It doesn't follow redirects, so it didn't work for four out of the six sites I tried. Ponyfish asks you to click links on the page that point to "articles that might change frequently." It then looks at the URLs and tries to come up with a simple filter, like "http://www.cthrall.com/blog/?p=*". Ponyfish comes up with some common denominator, and elements that are different become wildcards ("*"). <strong>Update:</strong> as you see from the comment, PonyFish has changed their site to follow HTTP redirects. I like their concept of generalizing the content URL.
May 3, 2006, 9:58 a.m.
I downloaded the <a href="http://www.cthrall.com/blog/DVD ISO">DVD ISO</a> for <a href="http://www.cthrall.com/blog/Fedora Core 5">Fedora Core 5</a> yesterday. For FC5, you need to choose "Other Linux 2.6.x kernel" when creating the VM. If you choose "RedHat Linux", it won't recognize the SCSI driver VMWare uses. I found <a href="http://blog.blaeke.info">Geert Blaeke's excellent blog</a> while trying to figure that out. I'd also forgotten that you need to point the CD-ROM device at the ISO file to install from an ISO image.
May 2, 2006, 3:05 p.m.
There was always one tech magazine that stood out for me. It somehow managed to comprehensively review all the hardware and software advancements that happened during its lifetime. It didn't compare and contrast the new printers, it covered the technical details of the new PostScript engine in NeXT, the difference between CISC and RISC, and it had articles by <a href="http://www.circuitcellar.com">Steve Ciarcia</a> and <a href="http://weblog.infoworld.com/udell">Jon Udell</a>. A couple years ago, I subscribed to the BYTE monthly e-mail. I just noticed the other day that <a href="http://www.byte.com">BYTE</a> is now availabe free online. Some of the names in the bylines I recognized, but the quality of the articles is the same solid reporting.
May 1, 2006, 10:09 a.m.
It looks like <a href="http://www.nytimes.com/2006/05/01/technology/01google.html?hp&ex=1146542400&en=76dac6927c261199&ei=5094&partner=GOOGLE">Google is not happy</a> (nytimes.com) with the search bar in IE7. From the article: <blockquote>The new browser includes a search box in the upper-right corner that is typically set up to send users to Microsoft's MSN search service. Google contends that this puts Microsoft in a position to unfairly grab Web traffic and advertising dollars from its competitors.</blockquote> I'm running IE 7 Beta 2, and the default search is Google. Also, you can click on a drop-down list on the search bar and change your default search. IE7 also has a "Find more providers..." link that is similar to Firefox's "Add engines..." link. IE7B2 defaulted to Google, and only Google. When I clicked "Find more providers..." and added Wikipedia, it also added "MSN Search." ;) Thoughts: <ul> <li>I wonder if Microsoft is just defaulting to Google until they release IE7. Then, they'll change the default to MSN Search.</li> <li>If Microsoft keeps the interface the same, with a "Find more providers..." page that contains Google and other providers, along with a description of what any search site needs to do to add a link on the search bar, I don't think Google has an argument.</li> </ul>
April 24, 2006, 9:28 p.m.
This is an example of a WordPress page, you could edit this to put information about yourself or your site so readers know where you are coming from. You can create as many pages like this one or sub-pages as you like and manage all of your content inside of WordPress.