automate the links/etc that handle non-standard date paths

[rjbs]

The 2011 (and 2012, when it's done) articles are in the format 2012/2012-12-01.html

In 2010, it was more like 2010/1/index.html

In 2003, you could say 2003/23rd/index.html

Sometimes there are symlinks, sometimes <meta http-equiv="refresh"...> and sometimes things just don't work quite right.

Let's make "20xx/20xx-yy-zz.html" standard, and set up permanent redirects for any plausible other form of entering a link. Then we can clean up all the stuff on disk.

I think the simplest way to do this might be a very thin Plack program, but something that emits lighttpd configuration might do in a pinch.

[rjbs]

This program could also deal with the RSS feed for whatever is most current, etc.

I'd like to keep the site mostly static content, but this seems like a pretty reasonable set of things to handle dynamically.

[n1vux]

Related, but simpler to solve (and currently visitor visible), so maybe this deserves a separate issue:

Archives pages http://perladvent.org/archives-Yd.html and ./archives-AZ.html are linking to 2010 uniformly as ./2010/dd but the pages are at ./2010/9/ , so there are (at least) 9 pages that appear to 404 from the Archive link.

Could be fixed with rewrite, symlink, meta redirect, or just editing the two Archives files to add q(0)x9 each.

(IIRC on the 2nd generation of Advent's PM.org WEBdav server, we had symlinks instead of rewrite rules. My peek at Archives today was prompted by PM.org discussion about moving that static service to "Grandfathered" twighlight status.)