On Sat, Nov 17, 2001 at 05:18:26PM -0700, Chris Tillman wrote: > > If you're talking about checking it _before_ downloading, I agree that > might be a nice enhancement, maybe just the first 100 lines or > something. Also a link to the README would be cool. But I think some > of these things are hard to do considering that the archives are > mirrored all over the world. > Unpacking .deb and grabbing the changelog/copyright files needs a fair amount of resources. It would be easy for someone to grab huge resources by requesting many changelogs. What is needed is to create a cache system so .deb are only unpacked once per package version.
Nothing is particularly difficult, but there are a number of cases to deal with: - some packages still use /usr/doc - /usr/share/<foo> may be a symlink to a different package - /usr/share/<foo>/<file> may be a symlink to a file either in the same directory or to a file in another package - not every package contains an upstream changelog file - for efficiency the extraction script should extract the changelog.gz changelog.debian.gz and copyright file all in one go. Additionally, the following issues need to be dealt with: One could simply present files for unstable, but that won't handle the case of packages that exist in stable, but not unstable. The script needs to be on a machine with a complete archive. It needs to be able to deal with different arches. To prevent an overly large, flat directory I would suggest creating a cache that mirrors the archive pool directory structure. Make it efficient. Any volunteers? -- James (Jay) Treacy [EMAIL PROTECTED]
