David Timms wrote:
At the moment, many .spec URLs that are hosted outside Fedora are
taken
down the moment or shortly after the process is complete. Hence it makes
it difficult to learn from the efforts of others in a package review in
Fedora because the .spec is no longer linked in the spec/SRPM urls
provided by the submitter.
The second issue is that browsing the cvs server is quite slow - mainly
at the point where the "folder" has the complete list of {3000} package
folders. This means we could immediately click through to the correct
folder in the cvs server, rather than trawling from the top level.
This is only a partial help since it's not easily discoverable by a new
user but it will keep you from having to browse the complete foleder
list on the cvs server::
http://cvs.fedoraproject.org/viewcvs/rpms/PKGNAME
Any comments, rejections to this idea ?
Like Rathann, I'm against this if it's a manual requirement but if it
were automated it would be useful. If you would be willing to
contribute some code, the cvs request needs to become something that is
entered into the package database. The bugzilla address will be entered
as part of this so that the cvs admin can continue to verify the
reviews. So it would be possible to automate adding a link to viewcvs
at the end of this process.
In the meantime, if you think it is a good idea, it would be
appreciated
for others to do this in any case - if that is allowed ?
If a packager wants to do this themselves they are welcome to do it.
===
Another idea along these lines would be to have a staging CVS server
that the reviewer would be required to commit the original and changes
to the spec under development. This could then be committed to the real
cvs server after acceptance. The submitter would still put his srpms on
his own url, not in the staging cvs server.
This would make it simpler to get a live diff on what the submitter's
changes have been, rather than viewing the spec as a whole and trying to
find the changes.
This could be useful but it would be a separate process for people to
learn and a separate cvs instance that would have to be maintained. I'm
not sure that we want to do this because of the overhead. At least,
we'd want to get the pkgdb's automated closing of package bugs written
first so we could see if we could make an automated opening as well.
(Note: This would require people, at minimum, to have signed the CLA
before submitting their first package.)
-Toshio