Friday, March 31, 2006

Blogger's AtomAPI authentication = SPLAT

Yep, I get it too. Thank you to the person who thought it would be an excellent idea to both reply to one of my Deepest Sender LJ community posts and email me directly to ask me what is wrong. As if I worked for Blogger, or something. My answer? Monitor the thread linked above.

(Written using the Blogger web interface)

Tuesday, March 28, 2006

The Most Annoying Thing About Web Development, Ever

Premature end of script headers

Damn you, CGI.

Monday, March 20, 2006

Re: Filtering out the Channel Noise

I think the best solution would be aggregator (similar to Planet Planet which powers Mozilla Planet) in which tagging and submission capacity is supported. Submissions allow for other non-RSS based systems to feed data into planet, such as Mozilla Wiki, and Bugzilla. By using dozens of tags or categories, as well as some meta tags (of course there would be a general tag as well for timeless who must have more mail than Santa Clause), one could have links to all new info in one place. What I’m thinking of is something that essentially is planet as we know it, and wiki rolled into one. Blog like tagging and posting capacity, but wiki style editing and managing. Submitters could add tags, and so can others.

Well, why not use Planet Planet? At least for the aggregation part, anyway. Most everything in the list is syndicatable via Atom/RSS — you then have a source which includes:

  • Planet Mozilla's feed (plus the stragglers)
  • Mozilla Wiki's Recent Changes feed
  • Bugzilla's bug list feeds (ctype=rss or ctype=atom)
  • Bonsai's CVS log feeds (e.g., http://bonsai.mozilla.org/cvslog.cgi?file=file&root=/cvsroot&ctype=rss)
  • Google Groups feeds (examples at http://groups.google.com/group/group_name/feeds)

Writing a feed generator for mailing lists is trivial (relatively speaking). It shouldn't be too difficult to generate a feed for the Mozillazine forums, although with their bandwidth concerns, it would probably be restricted to just that aggregator. I'm not exactly sure how syndicating the tinderbox output would look (or spaghetti, for that matter).

The part that interests me the most is how to syndicate IRC. My thinking is that you have a bot log the channel and also serve a webpage that allows one to select lines of the chat and submit them as a "post" with tags, etc. It would also make it easier to make bash.org quotes, but I digress. :)

Anyway, back to the system mentioned in the quote. I think the problem with my solution is the signal-to-noise ratio (with more noise than signal). The only options I see at the moment that would account for that are:

  1. Social human intervention - doing the bash.org thing and modding "posts" up or down.
  2. Simply not tagging/untagging the bad posts, and letting them percolate down to the bottom that way.

...It's amazing what I can think up when I'm not in school.

Re: Normalized data is for sissies

[...]and it's good to remember that relational databases themselves were a controversial idea.

I find it very funny that a lot of the ideas that I learned in Database Management and Information Retrieval were/are pretty contentious, even though they taught it as though it was gospel.