Re: Limiting piuparts results in DDPO
[ switching to -qa@ and piuparts-devel@ ]
I think we can assign severities to piuparts problems. for now 2 should
suffice:
* serious, e.g. "modifies/deletes files from an unrelated package"
* important, e.g. "leaves cruft in /var/log/$PACKAGE/ after purge"
For stable releases we usually disable the tests that result in
"important" bugs since these are not going to be fixed there anyway.
And serious ones in stable should be only a few :-)
If there is a serious failure (in a critical suite), ignore important ones.
Furthermore we should be able to rank the distributions and show only
the highest ranked failure - most likely the same failure shows up in
more than one suite.
E.g. if the package fails in sid, we don't really care if it also fails
after the upgrade from anywhere to sid. Or in the upgrade from sid to
experimental.
maybe
sid
testing
testing2sid
stable2testing
experimental
sid2experimental
stable
Well, that's only for failures: one should be enough. If there are no
failures we have states like "cannot be tested (any suite)" "waiting to
be tested (any suite)" and "all tested successfully (all suites)".
Andreas
still in -ENOTIME mode :-(
Reply to: