The characters <>" are now simply not allowed in URLs. Incidentally,
this also fixes formatting of some URLs within brackets, e.g.
<http://blicky.net/>.
The XML thing is supposed to be used for auto-completion, of course. The
current search implementation is very, very stupid and simple: Just a
prefix match on the man name, and simple detection of section stuff. I
suppose it'll suffice for now.
I couldn't get firefox to play nice with the search button styling, so
replaced it completely with an image. Its location still off by one
pixel in Firefox, though. :-(
Grotty outputs double-formatted characters for some (old) manual pages.
This fix ignores the second part of the double-formatting, thus making
things readable again.
This fixes the indexing of some packages. The most common error is to
have a doubly-gzip-compressed file with only one '.gz' extension, which
would previously not be detected.
This allows for a more dynamic nav bar without inserting insanely huge
HTML code in the page (as the previous version did in some cases) and
without having to contact the server again.
The 'intro' man page has around 1500 versions, and only generates a page
of ~52KiB (~9.5KiB after compression). The previous HTML version was
106KiB (~10.3KiB after compression). Page generation times have been
improved on the server side (by 50ms for the intro man page), but I've
no idea how significant the effect is of JS is. Feels fast enough,
though.
