Hacker News changed its RSS feed and the URL which used to return the top 20 stories is currently returning 81. This basically makes Hpricot shit its pants, although I'm sure part of the problem is the "quick and dirty hack" nature of the code. If you're interested in fixing it, the problem is probably just one literal hard-coded value, the URL to hit which returns the top 20 stories. I don't know if this is a temporary problem or a permanent one, and I don't care. Hacker News banned this domain, banned me from commenting, never actually said anything about it to me beforehand, have done weird things with their RSS feed before - the last time it took Hacker Newspaper down, the URL for the top 20 stories was returning 2MB of RSS - but most importantly of all, to me, it's hit a tipping point of unreadable dullness where it's not usually even worth the effort to load a web page any more, let alone the effort to debug the unpredictable, borderline incompetent RSS foibles of a NIH Web framework nobody else uses, except for a site which banned me, provides no API, and who I can safely therefore assume doesn't appreciate the effort.
As a chip-on-my-shoulder aside, Jim Weirich defeated Paul Graham's grand Arc Challenge with Ruby he'd written several years prior.
Anyway, my point is, fuck this shit. I know there are a lot of people who read Hacker Newspaper and thank me for it, and the Hacker News community produces staggering numbers of mashups, so if any of you want to fix it, it's open source and it's on GitHub. The version on the live server is somewhat diverged, but not in any way that matters for the purpose of debugging this, and I'll happily clean up the branches if necessary. I just don't have time to deal with this right now.
The last few times this happened, I added a quick workaround, or I just waited a day or two and Hacker News solved the problem at their end. This time, the problem is probably just one hard-coded literal value in a Python script. You don't even have to read Python to figure it out. You could also just have the Ruby check the RSS dump's file size, or maybe chop it up into more digestible pieces before using Hpricot.
If nobody steps up to fix it, I'll probably get around to it eventually, but it definitely won't be in the next couple days.