Re: RFC: Moving Java PRs to Gcc database

[Bryce McKinlay <bryce at albatross dot co dot nz>]

Tom,

I do have some concerns about moving the java bug database. Most
importantly, I presume that this will involve renumbering the existing
PR entries. This will mean that all the references to existing PR
entries in the mailing list archives, changelogs, and source code
comments will become invalid. Having working PR numbers available for
reference can potentially save a lot of time for anyone who comes
along later and tries to track down a regression or partially fixed
problem.

On the other hand I do agree that there are advantages to having a
combined database. There will more eyes on the PRs, useful when the
"bug" is really a generic gcc problem rather than a java one).

If you do move them over, then at the least I think that ALL the PRs
should be moved over, including the closed ones, since its often
useful to search through old resolved PRs to check a solution for a
problem that has reoccurred, for example. Also, it would be good if
your script would add some text like "Formerly PR libgcj/162" so that
the old PR numbers can still be searched for.

Ultimately, I'd really like to see a more powerful, comprehensive bug
tracking system that could ultimately cover the entire GNU tool chain.
For example, something which can track and manage dependencies between
bugs in different tools (ie: fixing gdb bug x depends on fixing gcc
bugs y and z, gcc bug z needs a fix for linker bug q, etc). I think
that a good bug management could be enormously valuable to GNU
developers and the project as a whole. GNATS is ok, its better than
nothing, but its very primitive compared to, say, bugzilla.

regards

  [ bryce ]