I replaced GeshiCodeTag's behavior with respect to unsupported languages so that it just falls back uses an HTML <code> block. These will be distinguishable from GeSHi-targeted <code> blocks because they won't have a language parameter associated with them. Be mindful of that. --Short Circuit 08:01, 30 January 2009 (UTC)
I already solved this bug. It's a very poor solution, it finds "<code>" and "</code>" indexes, replace them to "c0de", so re.sub() will ignore it. At script end "c0de" will be replaced to "code" again. Now the script can be 100% automatated, using Wikimedia API. (I think.) --Guga360 16:28, 30 January 2009 (UTC)
Please no "code" to "tt" change!
I dislike the new change of "fixing" <code>...</code> to <tt>...</tt>. The disabling of code tags is temporary, and using code tags (logigal markup) instead of tt tags (physical markup) is the Right Thing. I intentionally did not change code tags meant as inline code tags when changing code->lang (I did change inline code tags back when it seemed that the new meaning of code would prevail). --
- Sorry it was me; I realized your point and stopped me by myself before reading this, nonetheless I've already done little damages and will go fixing them as soon as I can. --ShinTakezou 15:46, 2 February 2009 (UTC)
- It's possible to protect this article from begin edited by bots? --User:Guga360
- Kinda sorta, yeah. I can currently offer varying levels of protection. Allow all normal access, disallow anonymous or disallow anyone but sysops. So I'd have to make it a sysops-only page, which stymies the page. It'd be better if your bot could be modified to attempt to detect code snippet boundaries and not modify content within them. --Short Circuit 23:35, 3 February 2009 (UTC)
This task is inadequately defined, since any tag would match <%s>.
should old.j be old.k ???