From: Reini U. <ru...@x-...> - 2004-06-24 13:48:33
|
John Cole schrieb: > Reini, > I'll bring up the latest cvs version on a test machine and try it out. > > Reini, your doing a whole lot of work right now ;-) and I'm sure everyone, > expecially me, appreciate it. When do you think things will settle down > enough for you to start working towards a new release? As soon as I have fixed the fatals and am on the same stability level as the release before. (say apache2, cgi, php5 must not crash. login must work ok. iis and better utf-8 support as bonus) Currently I'm investigating problems with application/xhtml+xml, which brings out some problematic XHTML code in the templates. Esp. javascript does not work as expected. (This is needed for some XHTML extensions, like MathML or SVG). Note that most if my new work isn't in CVS yet. It is not stable enough. Esp. my latest WikiGroup fixes which improved the mess a lot! > I'd love to see the code start to stabilize and not have new features > break so many things each time ;-) Even before the release, I'd like to see > the cvs tree tagged when you think things are stable for testing. > > There are so many nice features in the cvs tree, I think people would > really be amazed with the progress you've made on the system, if they could > see it without getting the cvs head ;-) -- Reini Urban http://xarch.tu-graz.ac.at/home/rurban/ |
From: Jon <d9...@na...> - 2004-06-24 20:23:04
|
On Thu, Jun 24, 2004 at 03:48:29PM +0200, Reini Urban wrote: > Currently I'm investigating problems with application/xhtml+xml, which > brings out some problematic XHTML code in the templates. > Esp. javascript does not work as expected. (This is needed for some > XHTML extensions, like MathML or SVG). http://diveintomark.org/archives/2003/01/14/eddies_in_the_spacetime_continuum1 There are still issues with XHTML 1.1, the way it should be served. It doesn't work with Internet Explorer, the error handling of xhtml+xml documents in browsers where it does work is quite weak, and it uses XML DOM instead of HTML DOM, though not really bad, confuses many javascripts. I have no idea how to develop for the web with the "current" (as of 3 years) recommended standards. If you find a feasible solution, mainly the IE problem, please tell me how you did it. -- Jon Åslund |
From: Reini U. <ru...@x-...> - 2004-06-25 09:11:09
|
Jon Åslund schrieb: > On Thu, Jun 24, 2004 at 03:48:29PM +0200, Reini Urban wrote: > >>Currently I'm investigating problems with application/xhtml+xml, which >>brings out some problematic XHTML code in the templates. >>Esp. javascript does not work as expected. (This is needed for some >>XHTML extensions, like MathML or SVG). > > > http://diveintomark.org/archives/2003/01/14/eddies_in_the_spacetime_continuum1 > > There are still issues with XHTML 1.1, the way it should be served. It > doesn't work with Internet Explorer, the error handling of xhtml+xml > documents in browsers where it does work is quite weak, and it uses > XML DOM instead of HTML DOM, though not really bad, confuses many > javascripts. I have no idea how to develop for the web with the > "current" (as of 3 years) recommended standards. If you find a > feasible solution, mainly the IE problem, please tell me how you did > it. I'm aware of the IE issues and more. We can output the headers according to the browser. The biggest problem is the inability to do document.write() (signout button, edit toolbar) with application/xhtml+xml or text/xml. http://www.w3.org/People/mimasa/test/xhtml/media-types/results -- Reini Urban http://xarch.tu-graz.ac.at/home/rurban/ |
From: Reini U. <ru...@x-...> - 2004-06-25 10:56:40
|
Reini Urban schrieb: > John Cole schrieb: >> Reini, >> I'll bring up the latest cvs version on a test machine and try it out. >> >> Reini, your doing a whole lot of work right now ;-) and I'm sure >> everyone, >> expecially me, appreciate it. When do you think things will settle down >> enough for you to start working towards a new release? > > > As soon as I have fixed the fatals and am on the same stability level as > the release before. > (say apache2, cgi, php5 must not crash. login must work ok. iis and > better utf-8 support as bonus) > > Currently I'm investigating problems with application/xhtml+xml, which > brings out some problematic XHTML code in the templates. This is done. php5 and CGI and apache2 works fine again. Apache2 with USE_PATH_INFO=true or auto works fine with "AcceptPathInfo on". Session crashes on overlarge session objects have also been fixed. Anybody knows the number of how large such a session object may be? Reading an overlarge object leads to immediate crashes. The most important remaining fatals are now: 1) ConvertOldMarkup() php crash on formatting the 2nd paragraph of OldTextFormattingRules, AnciennesR%E8glesDeFormatage, ... A complicated anchored regex which crashes in the 2nd loop. We really should check/sanify the allowed regexp's a bit, since admins and even users (InterWikiMap) can easily destroy them. 2) PageList memory exhaustion on large pagesets (e.g. PhpWikiDemo:en/AllPages) This is probably due to WikiDB caching on simple pagelist iterations, which shouldn't be done. -- Reini Urban http://phpwiki.sf.net/ |
From: Reini U. <ru...@x-...> - 2004-06-25 14:34:19
|
Reini Urban schrieb: > Reini Urban schrieb: >> John Cole schrieb: >>> Reini, >>> I'll bring up the latest cvs version on a test machine and try it out. >>> >>> Reini, your doing a whole lot of work right now ;-) and I'm sure >>> everyone, >>> expecially me, appreciate it. When do you think things will settle down >>> enough for you to start working towards a new release? >> >> >> >> As soon as I have fixed the fatals and am on the same stability level >> as the release before. >> (say apache2, cgi, php5 must not crash. login must work ok. iis and >> better utf-8 support as bonus) >> >> Currently I'm investigating problems with application/xhtml+xml, which >> brings out some problematic XHTML code in the templates. > > This is done. > php5 and CGI and apache2 works fine again. > Apache2 with USE_PATH_INFO=true or auto works fine with "AcceptPathInfo > on". > > Session crashes on overlarge session objects have also been fixed. > Anybody knows the number of how large such a session object may be? > Reading an overlarge object leads to immediate crashes. > > The most important remaining fatals are now: > > 1) ConvertOldMarkup() > php crash on formatting the 2nd paragraph of OldTextFormattingRules, > AnciennesR%E8glesDeFormatage, ... > A complicated anchored regex which crashes in the 2nd loop. > We really should check/sanify the allowed regexp's a bit, since admins > and even users (InterWikiMap) can easily destroy them. Status: added some debugging code to exclude certain regex to narrow the search. > 2) PageList memory exhaustion on large pagesets > (e.g. PhpWikiDemo:en/AllPages) > This is probably due to WikiDB caching on simple pagelist iterations, > which shouldn't be done. Status: fixed, needs testing. 3) php crash at some template expansion, if just logged in. Status: no guess in which expansion yet... I committed my current status with these remaining problem to CVS now. -- Reini Urban http://xarch.tu-graz.ac.at/home/rurban/ |
From: Reini U. <ru...@x-...> - 2004-06-27 08:37:37
|
Reini Urban schrieb: >> The most important remaining fatals are now: >> >> 1) ConvertOldMarkup() >> php crash on formatting the 2nd paragraph of OldTextFormattingRules, >> AnciennesR%E8glesDeFormatage, ... >> A complicated anchored regex which crashes in the 2nd loop. >> We really should check/sanify the allowed regexp's a bit, since admins >> and even users (InterWikiMap) can easily destroy them. This only crashes with Apache2. Very strange. Looks like another php bug caught us. > Status: added some debugging code to exclude certain regex to narrow the > search. -- Reini Urban http://xarch.tu-graz.ac.at/home/rurban/ |