On 6/1/07, Tim Starling
<tstarling(a)wikimedia.org> wrote:
Magnus Manske wrote:
On 5/31/07, Brion Vibber
<brion(a)wikimedia.org> wrote:
But do go ahead and mention specifics if you see
something awry.
Like this? :-)
http://de.wap.wikipedia.org/transcode.php?go=Felsberg+%28Hessen%29&seg=2
"Error: Invalid wiki syntax"
Perfectly normal and to be expected. The codebase has about as many bugs
as lines of code. It's the working articles that should surprise you.
Oh, you're writing a new parser for that? In that case, why not use my
wiki2xml stuff for the next generation? It was working OK last time I
checked (some month ago...) and usually breaks gracefully (dumping raw
wikitext in worsed case). It also comes, among other generators, with
plain-text output, which can be adapted to the output format of your
choice. Finally, it was surprisingly fast (the online demo sucks,
though, because it takes forever to get article and template texts on
the toolserver).
Writing a new parser is the wrong way to do it. Taking a TikiWiki parser
and hacking it until it partially works in MediaWiki is even worse. It's
rubbish code that has been dumped on me. I intend to get it working with
plain text and most images, but no tables or templates. Then we can think
about rewriting it. Maybe some sort of XSLT/PHP combination on the XHTML
output is the way to go, or maybe subclassing the mainline parser would be
better. But not an independent wikitext parser.
-- Tim Starling