I spent this morning further hacking on the axkit.org mailing list archive software (which I'm also using on the Photogeeks list archive).
The first component of this is an XML generator for ezmlm-idx indexes. It uses SAX, and it was kinda nice to get into XML a bit again, although I don't want to spend too long there :-)
After the index generator I have a whole bunch of XSLT files to re-thread the mailing list. This is quite nice, because XSLT is natively tree-based. The only complex bit comes when displaying orphaned threads in the month-view, but thanks to the guys on #axkit-dahut and Michael Kay's XSLT book I soon sorted that out.
All in all I think it looks fairly good. And a hell of a lot better than ezmlm-cgi.
Feel free to take it for a spin: http://axkit.org/archive/.
Re:gmane
Matts on 2005-08-13T16:56:09
Yes.
I have two complaints with most mailinglist archives:
Lacking i18n.
I send mail as UTF-8, because I like to be able to use better typography than straight quotes and such, I like being able to use real bullet points, I like being able to mix Greek and Latin characters freely, etc.
But most archivers either simply assume my mail is Latin1, or just obliviously deliver their pages as Latin1 without ever bothering to transcode/escape their input properly. The result is that on the web, my mails appear full of garbage – ironically, browsers in general have much better Unicode support than mailers.
Sigh.
Almost no archivers show you the Message-ID header of a message. This means when you’re perusing the archives before you subscribe to the list, and you want to reply to something, there’s no way to thread things properly because you can’t find out the message ID of what you’re replying to.
It would also be nice if there was a way to link to messages by their ID.
Re:
Matts on 2005-08-14T15:45:33
This does full i18n and it does it right even when the mailer gets it wrong. This isn't easy, but it's not rocket science either. Of course using AxKit and XML helps with this as it tends to make sure you get encoding just right or it falls flat.
Regarding Message-Id display, I could add that quite easily by tweaking the stylesheet.Re:
Aristotle on 2005-08-14T16:21:50
Great to know that you care.
:-) You are exactly right, doing things correctly isn’t trivial but no rocket science either – just takes care and some effort. I only recently saw another list archive get encodings right for the first time, and then another, so I’m hopeful for this class of application. I get discouraged sometimes by how many really, really basic things (i18n being just one of them) still don’t work right in so much software. We’re building castles on sand, it occasionally feels like. There’s so much ignorance, and education and better practices take so agonizingly long to become ubiquitous…
Anyway, enough lamentation.
