I've been working on a project for the last while that has, as one aspect, an RSS aggregator to bring together news from a bunch of different related website - nothing earth-shaking there. Problem is one of the sources I want to aggregate is an ezboard forum. The fact that ezboard doesn't have RSS feeds, however, put a bit of a damper on that.

A bit of Google searching made it quite clear that I was not the first person to want to do this, and while I am all for not reinventing the wheel when possible pretty much all I could find were lots of "ZOMG!!1!! I needs RSS from ezboard. PLZ spoon feed me!" forum posts and very little of actual use. In reply to these there were a couple of links to a dead service at http://www.mikeshea.net with the requisite "is gone OHNOES!!1!" replies. Eventually I found a post where someone rather usefully posted a link to the source on Mike Shea's site and I gave it a try.

While there was some wonderful Perl-y goodness to the script, particularly it's ability to deal with a whole bunch of other bulletin board services, the RSS generated was, to put it mildly, badly formed - the worst part for my application was a lack of pubDate tags so the items wouldn't slot in chronologically with the other feeds in my aggregator. Running the script also added about 20 lines of warnings in the error log every time it was executed. The script was all at once overkill and not very useful to me. But not completely useless. I retained the caching and some of the ezboard-specific screen scraping when I rewrote it, so credit where credit is due.

My script is entirely single-purpose, unlike Mike Shea's script it only works with ezboard forums, that's the downside. The upside is what it returns

1. is the right content type ('text/xml')
2. includes a lastBuildDate
3. includes a pubDate for each item
4. includes sticky threads, yet still sorts the thread items in the correct descending order by date/time, and, of course
5. runs without warnings or errors

How to use:

For the cache you'll need a subdirectory under the location of the script that is writable by your http server process.

The script expects three parameters:

1. title - arbitrary text describing the feed. This is what is going to end up inside the <title> tag of the RSS feed. Don't forget to URL encode the string.
2. url - the full URL to the forum you want to make a feed from, URL encoded, of course, and
3. description - arbitrary text  describing the feed, this will end up inside the <description> tag.

Example:

http://yourservername/andpath/ezrss.pl?title=Arbitrary+Title+for+Feed&url=http%3A%2F%2Fpxxx.ezboard.com%2Fmyforumpath&description=Arbitrary+Description+of+the+Specific+Forum

Source code: ezrss.txt

Some of you may have noticed I went unusually long without posting here, wasn't responding to messages on Facebook, mySpace, deviantART, etc. More of you probably noticed it took me unusually long to getting around to posting last week's Sin City pictures.

Pretty much every spare moment I had this past week went into writing a new upload interface for Gothic BC. The old one was slapped together over five years ago and was mediocre in 2002. I've been wanting to do a proper drag-and-drop control for a long time now and the opportunity was ripe since I needed to being the upload control in line with the new design anyway.

I "cheated" to a certain extent and bought a Java control to handle the drag-and-drop ("Hey, look, a wheel. Guess I don't need to reinvent that.") - what sucked up all my time for a week was *doing* something with the files that was better than the old upload. I'm pretty happy with what I've got going (especially a spiffy progress indicator while it builds the editing page). Much like when I first put the new gallery up there are still some problems: if you try to upload more than 75 or so photos in M$IE it takes forever to render the editing page

Nonetheless what I've got going now is as easy to use as Flickr, Facebook etc. which isn't too shabby since my budget is about five million times smaller.

( feature list and an annotated screen shot of the editing page )

I had a "lead developer" telling me that their application wouldn't work in a frame because if the address bar wasn't visible in the browser the application wouldn't work because it couldn't "see" the parameters in the URL. He meant "see" literally.

In a way I almost envy the guy. He lives in a magical world where servers have mysterious powers to visually read off URLs from the browsers address bar. That must be why MSIE shows the URL in pop-up windows even if the address bar is disabled (and you can have one guess what platform and technology this magician works with).

I asked him to explain how the GET request changes if the address bar isn't there. He said, "I don't want to argue with you."

The worst thing is the developer in question (or questionable developer, as the case may be) has lied to my client and I'm afraid he has convinced them that it was my fault and my only defence, sadly, it trying to give a technical explanation to someone is not in the least technically inclined.

I keep running into this scenario and I feel like I am defending evolution in Tennessee. It is depressing.

I was told this in all seriousness last Thursday:

I don't think it matters that our site doesn't look good in Firefox. I've looked at other big sites that don't look good in Firefox and we're no worse. Is it even possible to make a site that looks good in all browsers? I don't think it is.

All I could do was rub my neck and say, "No, actually it is."

*sigh*

One of the reasons I keep sinking time into Gothic BC is because I can use it as a sandbox. It's a good sized site with a lot of users (more traffic and users than most of the sites I am currently being paid to work on) where I am not hampered by corporate inertia. When inspiration strikes I can act on it.

In this case the inspiration was the Slide.com stuff I was looking at yesterday and Flickr. So, for those that might be interested, here is a sneak preview of where I'll be taking the gallery.

The 20,000' view of the technology here is this: the current Lotus Domino photo gallery is linked to a MySQL database by DECS and MyODBC, from there on it is run-of-the mill LAMP stuff. When it is done, and if time permits, the gory details of the Domino=>DECS=>MySQL=>PHP integration in my highly neglected and fantastically boring programming blog.

Tristan Risk in 3D as ASCII art. <-- Now in colour!

As before no animated GIFs were harmed in the making of this image, it's all plain text animated with CSS and JavaScript (that can't show on LJ directly). Once again, you'll have to follow the link to see this exercise in even more astounding uselessness than last time.

Rendering this should pretty much max out your CPU - this is about as far as I can go with this in this direction.

This is so completely pointless I'm at a loss to explain what possessed me to do it, nonetheless, I bring you:

Tristan Risk in 3D as ASCII art.

Note that this is not an animated GIF, this is plain text animated with CSS and JavaScript so I can't show it on LJ directly. You'll have to follow the link to see this exercise in astounding uselessness.

It works best if you sit back and squint a little.

I can't tell you how many times the situation as arisen where I have wanted to find a particular old blog entry to refer someone to and struggled to find it. Typically I'd employ one of three methods to find an old post:

• making a rough guess at the date and looking through the subject lines for the months near when I think I might have made the post,
• making a rough guess at the date and looking at the image folders on my server for a picture in the post, then going to the month in the archive and finding the post, or
• grepping my backup (I run a Perl script nightly that downloads my entries) to find the post.

Grepping is quickest but the results can be so ugly that the other two methods are often more effective. So, today I decided to write a search utility that gives nice, clean results. Finding things in my blog is now a snap. And as it gets bigger (there are already almost 3,000 entries) this will get more useful.

There is also another level of utility to it as well. More than just being a useful tool, it was an exercise in using Perl to pull together elements from my Lotus Domino server, a third-party website, and the file system in a nice, seamless way - which is exactly what I am going to need to do for the leverage the work already done with the artofadornment store for selling prints. There is a plan, it all fits together eventually.