On 07/17/12 22:34, Tim Foster wrote:
On 07/18/12 03:59 PM, Danek Duvall wrote:
Tim Foster wrote:
I've another tiny code review here, causing us to skip the test which
requires the test machine to have a particular locale installed,
rather than
raising an error if the locale is not present.
Looks good to me.
Thanks for taking a look.
Is there anything we can do to the hudson setup to
complain when tests are skipped?
I think we could. I dug around what I think is our hudson test
script, "pkg-build.bash" this afternoon. It returns an non-zero exit
code when it discovers the string "BASELINE" in the test log,
expecting to find output, like:
BASELINE MISMATCH: The following results didn't match the baseline.
(grepping for that text would be more bulletproof, but anyway...)
If it also checked that any lines in the log which matched the
expression "s - skipped" also matched the expression "s - skipped 0
tests", then we'd be able to fail on skipped tests too.
(I don't know if Hudson return a third "pass, with warnings" state)
I didn't want to screw about with hudson without coordinating with
Dave first though. Dave, should I go ahead and modify
/rpool/builds/pkg-build.bash on hudson-qe, or does that script live
elsewhere?
cheers,
tim
Tim,
You can add that in if you'd like. I made a quick
pkg-build.bash.20120719 copy to be on the safe side. The plan is to put
that script (and maybe a couple other hudson related things) in the gate
soon and give it a real code review.
Dave
--
Dave Vadovszki
(303)272-2686
500 Eldorado Blvd., Bldg.5
Broomfield, CO US 80021
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
