Date
1 - 2 of 2
[DISCUSS] Travis builds and PR testing
sjudeng <sju...@...>
Now that #251 has been merged tests in all modules except hbase-098 and
cql are required to pass for a successful Travis build. However since
merging there have been more unexpected build failures than were
observed during testing as part of that PR. The open #279 should resolve some of
these failures but it's likely there will still be some unexpected
failures of Travis jobs.
Are there concerns/comments regarding continuing to require the additional modules to pass for a successful Travis build (and PR merge)? In the case of unexpected failures committers can login to Travis and restart the failed job(s) without restarting the entire build (see attached). While this is not automatic I think it's still easier than requiring PR reviewers to run tests locally or request test summaries from contributors and it also keeps these test checks more public.
Are there concerns/comments regarding continuing to require the additional modules to pass for a successful Travis build (and PR merge)? In the case of unexpected failures committers can login to Travis and restart the failed job(s) without restarting the entire build (see attached). While this is not automatic I think it's still easier than requiring PR reviewers to run tests locally or request test summaries from contributors and it also keeps these test checks more public.
Ted Wilmes <twi...@...>
I'm okay with continuing to require the additional modules. You guys have made a lot of progress on test stability and this will keep that at the forefront.
Thanks,
Ted
On Sunday, May 21, 2017 at 9:15:24 PM UTC-5, sjudeng wrote:
Now that #251 has been merged tests in all modules except hbase-098 and cql are required to pass for a successful Travis build. However since merging there have been more unexpected build failures than were observed during testing as part of that PR. The open #279 should resolve some of these failures but it's likely there will still be some unexpected failures of Travis jobs.
Are there concerns/comments regarding continuing to require the additional modules to pass for a successful Travis build (and PR merge)? In the case of unexpected failures committers can login to Travis and restart the failed job(s) without restarting the entire build (see attached). While this is not automatic I think it's still easier than requiring PR reviewers to run tests locally or request test summaries from contributors and it also keeps these test checks more public.