From: Joseph B. <jb...@um...> - 2004-09-22 11:42:38
|
If the page is dynamically generated from the readme files, then it might be best to not access CVS in the script (for speed concerns). For that we'd probably need a local checkout of the CVS components branch. How feasible is it to have a copy checked out? Does SourceForge allow for that much space? The file release information can't be pulled from CVS. Are the file releses held in a directory heirarchy so that we could figure out if a component has a release? Or is parsing through the releases page look like the safest way? Even if it isn't possible to make the page dynamic (without slowing down the end-user experience too much), I don't think that a 24 hour delay to generate a static page via a cron job would be too bad. At least it would take some of the burden off of you. Aside from generating the comlist.php page, how do the pages currently under each components "website" directory end up in the live http://wxcode.sourceforge.net/components/ComponentX/ directory where the user can see them? Since it's possible to browse to http://wxcode.sourceforge.net/components/ComponentX/CVS/, I'm guessing that it's a cvs checkout/update which could hopefully be automated as well. |