Since builds on master currently fail the valgrind step, here's my
proposal for the temporary way forward:

1. request that the failure threshold on the valgrind step be bumped up.
That way all of the reports remain visible until the team decides how to
deal with this (see other emails).

2. If you have a change to master that is failing and you need to get it
through you can rebase your change on this revert from Phil:

https://gerrit.iotivity.org/gerrit/#/c/22301/

That will allow the change to build, since the test environment will not
save one memcheck file which contributes 8 fails, enough to put the
count back under the limit.  the Jenkins log for your build will now not
show all of the memchecks, unlike the note in #1, but should pass if
nothing else is wrong.

===

Longer term we have to decide whether the memcheck results are worth
having contribute to the jenkins vote.  I've seen a fair bit of comment
on the Internet that suggest gtest and valgrind don't make a really good
combination, although there are certainly others using them in concert
as we're trying to do.

If they are, we will have to write suppressions for things we decide are
not errors important enough to count towards a fail.  There's a way to
do that, but I have no personal experience with it. Basically, you
examine everything, decide what you want to suppress, and then, using a
text output file instead of an xml one:

Run valgrind as usual, but with the extra option --gen-suppressions=all.
This tells valgrind to print a suppression after every error it finds

You then clean that up and using your list, add the ones you want to the
suppression file (the current one lives in
iotivity/tools/valgrind/iotivity.supp)

_______________________________________________
iotivity-dev mailing list
[email protected]
https://lists.iotivity.org/mailman/listinfo/iotivity-dev

Reply via email to