Hi, Matt's patch for packages.d.o generation scripts seems to have worked fine, with a couple of tweaks.
In the meantime I've been told that Jason didn't allow NFS mounting (reason not stated) so I went ahead and made a script[1] to get the Contents-* files from a nearby mirror site. Interestingly enough, we should have had such a script for that already, in order for the CGI[2] to search over non-US files. So, that's all set up now, the script will run daily and update the files using wget -N (I thought that was better than weekly even though they get regenerated weekly, because they might change unexpectedly for whatever reason[3]). I've modified the CGI script to grep over the non-US file for the selected distribution, it seems to be working fine. During testing the CGI, however, I discovered that the simple patch to make links to per-package web pages wasn't quite enough -- it didn't handle the case where two packages own a file, and it didn't handle packages in contrib and non-free. I've fixed the latter (simple) and kludged around the former; I suspect someone will have a patch to fix that (and have the script make coffee, too ;) within hours :) Note that if the old mirror on master:/debian2/ was removed we could see if anything still uses it almost instantly... and get loads of free space for all that warez and mp3s^W^W^W^W^W^W too, of course :) [1] master.debian.org:/org/packages.debian.org/contents/refetch [2] master.debian.org:/org/cgi.debian.org/cgi-bin/search_contents.pl [3] Bug#82551: Contents files in stable contain weird binary characters :) -- Digital Electronic Being Intended for Assassination and Nullification