Thanks! Related to codecov, does anyone have any thoughts on changing codecov.io settings so the GitHub actions are only in the "informational" mode:
https://docs.codecov.io/docs/commit-status#informational This way if there are any missing lines of code, it won't fail the check. I think we would still get the inline notices saying that a line has missing coverage, but a PR will still show the check as passing, and commits will also show a pass. Alternatively we could set the "only_pulls" setting: https://docs.codecov.io/docs/commit-status#only_pulls This way PR codecov checks can still fail the check if not sufficiently covered, but if we decide that's fine and merge it then it won't cause the merged commit to fail the check. On 4/27/21 1:20 PM, Beckerle, Mike wrote: > Created https://issues.apache.org/jira/browse/DAFFODIL-2509 > about adding these coverage exceptions uniformly for all the Assert.xyzzy > where it is applicable. > ________________________________ > From: Steve Lawrence <slawre...@apache.org> > Sent: Tuesday, April 27, 2021 12:45 PM > To: dev@daffodil.apache.org <dev@daffodil.apache.org> > Subject: Re: codecov - annotation to suppress false positives? > > We use the sbt-scoverage plugin for generating coverage measurements > before sending them to codecov.io for display. It looks like this does > have a way exclude packages and sections of code: > > https://github.com/scoverage/sbt-scoverage#exclude-classes-and-packages > > So we could maybe do something like: > > foo match { > case real1 => ... > case real2 => ... r > // $COVERAGE-OFF > case thingy => { > Assert.invariantFailed(".....msg...") > } > // $COVERAGE-ON > } > > To exclude the entire case that should never be hit? > > > On 4/27/21 10:29 AM, Beckerle, Mike wrote: >> We have assertions like: >> >> foo match { >> ... real cases .... >> case thingy => Assert.invariantFailed(".....msg...") >> } >> >> The same thing can happen with if-then-else logic obviously where you make a >> decision, and some paths >> through the logic can't occur. >> >> These always get marked as non-covered, because by nature they're never >> supposed >> to happen. >> >> Is there a structured comment or some way to tell codecov that this is ok, >> and >> not to issue a warning about this line? >> >> Mike Beckerle | Principal Engineer >> >> mbecke...@owlcyberdefense.com <mailto:bhum...@owlcyberdefense.com> >> >> P +1-781-330-0412 >> > >
