First, I copied my favicon.ico to my S3 bucket. It’s just a 16×16 png file that I renamed.

In my settings file I made sure the MEDIA_URL was set to my S3 URL.

Then I made sure these things were imported in my urls.py

from django.conf import settings
from django.views.generic.simple import redirect_to

Lastly, I added this rule to my urls.py

(r'^favicon.ico$', redirect_to, {'url':settings.MEDIA_URL+'favicon.ico'}),

That’s it!

Before the update, the server was running a build of Gentoo from 2005. It was also running PHP4, a very old MySQL version, and a very old Apache 2. Now I have it running a new Ubuntu, lighttpd, PHP5 and new MySQL. All in all, there isn’t a noticeable change in how fast the server loads pages. However, there is a huge increase in security, ease of administration. I’m also sure the server can now perform much better under a heavy load than before, although the old build did handle quite a Digging without a hiccup.

Also, before performing the upgraded, I assumed it was a project that would take a day or two. Mostly thanks to the awesomeness of Ubuntu and Linode, porting all the web apps over took only a handful of hours. Completely rebuilding a web server from scratch and restoring all existing services in just a few hours, without any data loss, is pretty amazing. The only hitch I ran into was when I couldn’t finish a MySQL dump because the partition with the old build ran out of space.

Despite all this good news, there was one hitch. You see, this blog actually started running on a perl-based CMS known as blosxom. It’s still around, but nowadays it has a hard time competing with the likes of WordPress or Drupal. Anyway, Blosxom had a very elegant URL structure. When I originally switched to WordPress, I had to create an htaccess file and use mod_rewrite to keep that existing URL structure and not break every link to my blog.

Now that I have switched from apache2 to lighttpd, htaccess is no more. All the clean URLs for this blog are gone in favor of the ugly WordPress defaults. Every link to a specific article on this blog, including all the ones from my recent Digging, are broken. I could, if I wanted to, recreate the functionality of the mod_rewrite htaccess in a lighttpd configuration, but I just don’t care. I added a configuration to set the WordPress 404s to work, and that’s it.

This does bring up an interesting point though. URLs are largely determined by the web application and the web server working in tandem. If you want to move your site to a different platform, its a huge pain in the ass to keep the same URLs. This problem is largely unsolved. You’re pretty much either going to write lots of regex rewrite instructions to always keep your old schema, or you’re going to make broken links all across the web.

All in all, I think its best to just change and move on. The sooner you give up on the old, the sooner links will right themselves. If anyone out there has permalinks to any stuff on this site, feel free to update them.

I’ve learned that in order to get Dugg up, the headline is everything. Because it is my article, I actually bothered to read comments about it on multiple sites. The vast majority of people did not read the article. That may be because it was a wall of text, but I doubt it. People might read your headline, click your link, and skim your article, but that’s it. All you need to do to get a few thousand hits from a site like Digg is the headline.

Some people, including myself, attach some stigma to the idea of submitting content you have created to sites like Digg. I have abandoned this stigma. Even if you submit your own thing, the most that gives it is one vote. As of me typing this, my story has 1161 Diggs. No one has any motive for increasing exposure to my blog, with no advertisements on it, other than myself. Apparently 1160 people think my blog post is worth clicking. I do not feel bad for Digging it once. This is a stigma that should go away.

The most important thing I’ve learned is that sites being unable to handle large amounts of traffic are just poorly administered. Sure, my blog completely lacks graphics, and that helps. However, it also runs WordPress, which is infamous for not being able to withstand heavy traffic. Because I have the server properly configured, it was able to perform perfectly under heavy load. I don’t even do any weird caching or anything. It’s just a basic LAMP setup. I think what helps is that we pay $80 a month for hosting from Linode. Most people have these shared hosts for five bucks a month. Listen people, that isn’t going to work if you actually get significant traffic. You get what you pay for.

Also interesting was that I got a significant amount of non-Digg traffic. It should have been obvious, but I didn’t expect it. If thousands of people see and like my story on Digg, obviously they are going to link to it from other sites such as Reddit, but also on relevant news sites. Quite a few Linux or technology news sites around the world picked up the story, or it showed up in their news feed sidebars. Digg might be able to provide the initial exposure for your link, but the link will quickly fall off to the second page. Meanwhile, other sites that pick up your link will have you on the front page for quite awhile longer.

All in all, I think this was a valuable learning experience. I’ll probably be Digging a lot more of the content I create, especially if I have a feeling it is going to be popular. It helps a lot that my fears of having our server cave in have been eliminated.

I first attempted to find a free program I could download to run this on my own web server. That doesn’t seem to exist in any good and usable form. So I tried to write my own. This proved to be more difficult than I first realized. Since nobody seemed to format their feeds in a standard way, it was turning out to be a daunting adventure in XML parsing.

So in the end I combined two different services to perform the task. The first service is FeedJumbler which does a serviceable job of combining feeds. My only complaint is that it destroys a great deal of the extra data in the feed items. The other service is, of course, FeedBurner. It does a good job of taking the jumbled feed and making it a feed that is useable. It also adds in my del.icio.us feed and my flickr feed without killing the extra data.

I hope one day soon I will be able to use FeedBurner for this entire operation. So many people want to combine many RSS feeds into one, but most services that perform this function are meant for people who read rss rather than those who publish it. Not only that, but so few of them actually parse all of the XML properly causing the inevitable loss of useful tags.

I know from coding a solution why this is. RSS sucks. From the user perspective it’s great because programs that automatically create RSS, like WordPress, and programs that read RSS, like Firefox, already exist and work great. From a developer’s perspective it is a nightmare. I was not able to find a single RSS library in existence that actually handles anything that might be in a feed. No matter what tags from whatever XML namespaces appear in a feed, it should not be ignored or removed. People obviously are pretty good at coding software that writes RSS which is chock-full of semantic web happiness, but nobody has written anything that can make use of all of it let alone just hold onto it. Every library took the RSS data and destroyed at least some of it to make it easier for developers to handle it programatically. Then they made it incredibly difficult to re-write a new RSS feed.

I’m obviously a big fan of the uses of RSS, but not until we have intelligent RSS handling libraries which can actually make use of the full potential of RSS will we see more new and exciting things like podcasting.

Secondly, you might notice some nice changes to the CSS here. I’ve been working on a new theme for the podcast site lately. While I was doing that I learned some stuff about web 2.0 style and typography. That made me realize the fonts on this site were small and ugly. You should notice I’ve already made changes to the style that should make this page more pleasant to look at.

Another change you will notice is that I have removed the automatic del.icio.us cross posts from the front page and the RSS feed. You can still check them out by going to the actual del.icio.us category however. I liked the fact that I could backup my del.icio.us links to my blog, but I didn’t like the clutter they created. I finally found a technological solution to create a happy medium. I also have changed the arrangement of links on the right side of the page.
If I had a penny for every time a content creator on the web asked forgiveness for not creating content reguarly enough I could buy and sell the Sultan of Brunai. So I’ll let the sultan breath a sigh of relief and skip it. But I would just like to let you know that more awesome stuff will be appearing here soon. I’m doing a pretty good job of getting my affairs in order, and in the coming spring and summer months I have more energy to get things done.

Carry on.

When I saw this fps implemented using the canvas element last week I knew gold had been struck. With the canvas you have the ability to create a real-time drawing surface with JavaScript. With the XMLHttpRequest you have the ability to get and send data to a web application which can read and write to a database. If you add these two things together you can make a massive multiplayer online game in a web browser that requires very little loading time, no installation and no plugins. This is the hotness. If games like Kingdom of Loathing pick up on this technology they can do some really revolutionary stuff.

As a proof of concept for this idea I have created a terrible terrible demo. It’s really simple and almost embarassing, but that’s what you get with less than an hour of work. I only tested it in Firefox 1.5, but it should theoretically work in any browser that supports XMLHttpRequest and canvas. There are two links and a canvas. If you click a link, the square with the coordinates it represents will turn black. This isn’t anything new, except that behind the scenes those coordinates are being read from XML data. If you replace that static XML data with a web application and a relational database then you’re playing with power.

Maybe the demo will be better by the time you read this. I’m going to keep working on it as a platform for creating games that utilize grids. Tile-laying games like Tigris and Euphrates and strategy games like Dungeons and Dragons combat are top contenders. All that we have to do now is expand the canvas to have real 3d drawing power with the help of the GPU.

What are the long term implications of this technology? The most obvious thing I see is another nail in the coffin of Microsoft. Looking for the reason? PC Gaming is one of the main reasons pepople still run Windows. If all the software is web-based then Firefox becomes the OS and Linux/OSX/Windows doesn’t matter. If we can make a 3d, or even a 2d MMO that works in Firefox regardless of platform we can make a killing. Even people running old versions of BSD would be able to play the game just fine. There are more PCs than consoles out there. Firefox’s popularity is growing daily. A quality game built on this platform could easily overtake the great World of Warcraft because the barrier to entry is so small. To get into most MMOs you have to install software and have sufficient 3d hardware. Signing up for this MMO would be no harder than registering for a forum on the web.

I’m going to talk about this tonight on GeekNights.

I remember back in the 90′s there were JavaScript games. Java and Flash were big and slow to download over dial-up connections. Yet people wanted to play games in the web browser. People created a bunch of CGI and JavaScript based games using form elements and gifs to create poor renditions of classic turn based games. Mostly card games IIRC. But as technology advanced we got Director, Flash and Java to provide the browser embedded games we desired.

Recently there have been advances in web technology like AJAX, SVG, XmlHTTPRequest that allow web pages to be more dynamic than ever. Without using iframes we can have pages that both update information in a database and update their display based on information from the database without reloading. I think it’s freaking obvious that we should make a game based on that architecture. It will load fast, look cool, not require plug-ins, work in any new browser, allow multi-player and just be fun.

My current vision is to recreate versions of European board games such as Puerto Rico, Tigris and Euphrates, St. Petersburg, Carcassone, etc. using this platform. Much like the version of Settlers of Catan made with Java, this would be a huge hit and a great way to kill shit tons of time.

Now, there is more to it than in-browser games that don’t require a plug-in. ANY modern browser will run the games. That could potentially include browsers on cell phones, Palm devices, Blackberries, etc. Imagine if instead of paying your cellular carrier money to download a shit version of Tetris because they disabled phone functionality you can instead just visit a page and play board games with your friends for your standard data charge. And different style sheets for different media will enable the game to have a proper interface despite your chosen browser, device and network speed. The day when 5 people can play the same game of Puerto Rico on a phone, palm, blackberry, PC and PSP in different parts of the world will be a very good day.

I considered some co-location, but I think I’ll have to set that as a long term goal. I need money now for a security deposit on an apartment, not for another computer. Luckily I found Linode. Basically what they do is utilize the user mode Linux technology to turn one giant server into 20 or so normal servers. Then they sell each of the 20 servers to people like me. UML is really cool because what it does is virtually turn a single physical hardware computer into as many fully functional Linux machines as you want. It’s kind of like VMware, but for Linux only and 1000 times more powerful.

Originally I was hesitant about getting it. I was afraid it wouldn’t be powerful enough. But it was the only place that was providing complete root control of a Linux box for a reasonable price. So I took it and I can say so far I am very pleased. The web interface is especially nice. I can reboot the computer all i want and re-image with any distribution from their image server at the push of a button. I’m running Gentoo as always and administering it the way I like. And despite being a bit low on RAM I’m still pulling through quite nicely.

One thing I would especially like to note is that user mode Linux is free like all open source stuff. You can turn your computer into 50 virtual Linux boxes if you want at no cost. No other OS I know of has a freely available and high quality way to run multiple copies of itself on a single piece of hardware simultaneously. I know BSD has the jails, but its not the same. UML rules, this site will be hosted for a long while. Hooray and hooray.

Anyway, blog spammers are kind of easy to get rid of. You can set up all kinds of stuff like making the user look at an image and then type the number into a box in order to post a message. Or you can do it Engadget style and make users confirm by e-mail. But this blog is powered by a small perl script called blosxom. And the posting of comments is done via a smaller plugin perl script called “comments”. So instead of re-doing my entire blog in a different system I modified the comments plugin to not accept any comment in which the checkbox isn’t checked. My friend Steve came up with this temporary hacky fix. Originally I thought it was a hacky fix, but I eventually caved in and used it because the blog spam was getting exponentially worse. When I figure out a better solution I’ll do it.

You also might have noticed I upgraded the site in a few ways. I added a way to go through archives at the top there. I didn’t realize before there was no way to go back in my blog, so I got a blosxom extension for archiving that is extremely simple. It’s fairly obvious. Also, I moved the RSS link to the top from the bottom where it previously resided. With Firefox 1 RSS is become more popular, so I figure I should put it in a more prominent spot. I also added the appropriate tag in the header of my pages so the RSS feed will appear in the bottom right of firefox and other things that pay attention to that will see it.

Lastly I upgraded the creative commons license to version 2.0 instead of 1.0. I also corrected a few issues to make sure I still validate at XHTML1.0 Strict and the CSS still validates also. Oh yeah, reading my blog in Internet Explorer, which is perfectly standard XHTML and CSS, is worse than ever. Get Firefox, what is wrong with you?! Lastly, I’m considering adding a new rectangle thingy at the bottom to proclaim the blog is powered by Blosxom. In fact, I’m doing that this second, so anyone reading this will probably see it.

That should take care of the web design tasks of this blog for quite awhile. I have a predicament though, that in just a little over 10 weeks I graduate college. After that I’m on co-op. And unless I find a co-op in the Rochester area, I’m blowing this popsicle stand for good. That means the server isn’t going to stay in the apartment much longer. So I have to find hosting or get a new place with a domain and a net connection. Hopefully I’ll be paid enough money and I can co-locate the server cheaply. Maybe I’ll get www.apreche.net or www.gourmetgeek.com or something. Yeah… Anyway, watch out for that in the future. I still don’t know what I’ll do about my e-mail address.