On Sat, 8 Nov 2003 23:36:43 -0500, Lamar Owen wrote: > Multiple repositories are healthy, IMO. I don't buy that. Package foo in repo 1 doesn't include the fine HTML manual. Package foo in repo 2 includes docs, but they are incomplete (missing the diagrams) due to incomplete build dependencies. Package foo in repo 3 is complete, but is configured with different parameters than the other two packages. Package foo in repo 4 does not disable optional features in the src.rpm, so when rebuilt, it doesn't result in reproducible builds and depends on what software is installed. Package foo in repo 5 is fine, but its -devel sub-package has incomplete dependencies and causes trouble upon building dependent software. Package foo in repo 6 is complete, but the packager has not had the time to add an initscript, which makes the software less easy to install/use. Package foo in repo 7 is marked relocatable when in fact it doesn't work when installed into a different location. However, it contains a few bug-fixes from CVS. Package foo in repo 8 depends on maintainer-specific helper tools before it could be rebuilt or installed. That's our scenario. > Choice is good. That's a superficial argument. If you had written "diversity is good", that would be input for further discussion. But "choice" does not imply "diversity". > But with choice > comes responsibility: if you can't handle the responsibility of choosing > between RPM repos, well, maybe you need to go back to Windows. :-) Multiple repos create competition. Competition in the area of creating RPM packages is not good, because package bugs lower the quality of a package. > I'm not sure it would be great to see all these repos cooperate. But sure! It would be great help if e.g. some workload would be taken off Matthias Saou's shoulders [as he doesn't scale] and he could spend time on looking into non-packaged software instead or helping to develop tools which aid the packager. Of course, if that is not in his interest, oh well,... :) > Maybe their differences are useful to some people. Provided that there are meant to be differences, that fits into the "Fedora Alternatives" concept. However, if the only differences are package bugs or different spec file formatting issues, that's unfortunate. Ultimately, in a community project, there would be enough redundant resources, so a package could be maintained by someone else while the original packager is on vacation and a bug-fix or security-fix is necessary. > It might be useful; it might not. > But if I am going to put up a repository and pay for the bandwidth, then I > get to set the rules for my repository. [no comment] --
Attachment:
pgpKzB8Xe3znn.pgp
Description: PGP signature
