I agree with your concerns Leonard. But I think it is necessary to
come up with a solution for the case where someone makes a resonable
change to an interface which is used by many files in the kernel,
encompassing perhaps 40 or 50 maintainers who would need to be
contacted. What do you suppose be done in such a case?
It largely depends on the type and magnitude of the change. In the case of the
recent interface changes to PCI configuration, installing an entirely new
interface and deprecating the old without deleting it was just the right thing.
Individual drivers did not need to be updated instantly, so the author sent
patches privately to maintainers who could then update their drivers.
For smaller changes to an interface there currently isn't any option except for
the author of the change to update the individual files, and then for other
maintainers to send an additional patch if they aren't happy with the way the
change looks.
However, there is a way to improve the interaction. I've long thought that
there should be a separate mailing list for maintainers, not to take the
discussions from linux-kernel private, but so that there was a way to tell all
the maintainers that a particular issue needed to be addressed. Announcements
to linux-kernel are also a good idea, but there's no guarantee that with the
volume and occasional mailing list problems that all the maintainers will see
the announcement. If a change were announced on a maintainers list and people
given a little time to respond, maintainers could send in the appropriate
updates.
It's always best if the maintainer is given a chance to integrate any necessary
changes. Many times a given change has implications that are not understood
by the original author, and while it may work, it leads to accumulated cruft
if the maintainer doesn't integrate it properly.
Leonard
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu