On Wed, Jun 11, 2003 at 06:05:24PM -0500, Wolfgang Bangerth wrote:
Wolfgang, do you think it would be worthwhile to add a flag to PRs
that
are good candidates for regression hunts?
I haven't followed this too closely lately, but I had the impression
that
knowing which patch fixed something (unfortunately) did not imply
that the
bug was fixed soon. But that's a question those have to answer who
_do_ fix
bugs, not us bugzilla people. If they say that's useful information,
then
let's do it.
I kept track of this for a few months; regressions whose introductions
are reported aren't fixed sooner, but the people who fix them say the
information is (usually/often) useful. In several cases it got the
right people to take a close look at PRs and notice that the test case
was not valid, and that's also worthwhile.
Apart from that, if you have spare cycles: there are ~150
regressions, most
have small testcases these days, and most have not seen a regression
hunt.
Looking through PRs to figure out which ones might be good candidates
can be very time-consuming.