I've been wondering about an improvement to wikiTeX.
AFAIKS, there is no way, given the raw wiki code to locate an image generated from a section of wikitext on the server.
This would seem to be desirable, so third parties who do an XML export of wiki content can display the wikitex sections
At the moment I'm working on an experimental project (that will be properly anounced soon) that uses javascript to parse wikitext and generate the HTML on the client. Of course, I have no way to generate images so being able to locate the already generated ones on the main server is necessary.
On wikisophia.org's examples the math image is: /extensions/wikitex/tmp/2d129aef84955396d6b7a75c141ea41a.gif
I think some kind of namespace for automatically generated images would be helpful, for example:
/images/wikitex/Wikitex/math1.gif (where Wikitex is the name of the article, and math1 indicates the first use of math wikitext on that article)
-- Jim
On Oct 8, 2004, at 2:21 AM, Jim Higson wrote:
AFAIKS, there is no way, given the raw wiki code to locate an image generated from a section of wikitext on the server.
I'm fairly certain the filename is a hash of the text (probably after some limited processing).
I think some kind of namespace for automatically generated images would be helpful, for example:
/images/wikitex/Wikitex/math1.gif (where Wikitex is the name of the article, and math1 indicates the first use of math wikitext on that article)
This would make it impossible to locate an image based only on the text, since you would also need to know the context (which page it's on and what other TeX areas are on it).
It also would mean you'd have to separately rasterize the same TeX fragments when they appear on different pages (waste of CPU time and disk space) and regenerate all TeX fragments on a page when you change the order (ditto only more so). Also this would create caching problems, since a cached TeX image would become incorrect when the order changed.
-- brion vibber (brion @ pobox.com)
Brion Vibber wrote:
On Oct 8, 2004, at 2:21 AM, Jim Higson wrote:
AFAIKS, there is no way, given the raw wiki code to locate an image generated from a section of wikitext on the server.
I'm fairly certain the filename is a hash of the text (probably after some limited processing).
Ah, in that case the image URL is probably determinable from the wikitext anyway. Unless there are collisions I suppose, but the hash is so long I doubt that'll be a problem.
Thanks for clearing that up.
Ah, in that case the image URL is probably determinable from the wikitext anyway.
Raw MD5 as a matter of fact, Jim.
-- Peter Danenberg . wikisophia.org ..:
wikitech-l@lists.wikimedia.org