> On Jan 31, 2018, at 8:41 AM, Jeff Squyres (jsquyres) <[email protected]> > wrote: > > On Jan 31, 2018, at 11:33 AM, [email protected] wrote: >> >> If CI takes 30 min, then not a problem - when CI takes 6 hours (as it >> sometimes does), then that’s a different story. > > Fair point; that's why I experimented with (and accidentally left enabled) > only having the 2 pretty-much-immediate CI checks (email checker and > signed-off-by checker). > > We have definitely seen unreliable CI hang for hours (or days... or even get > abandoned when a CI server is reset). So it's understandable that sometimes > people merge before waiting for CI to complete. > > But I think the central question here is: do we want to leave it set as it is > right now: > > 1. you *must* make a PR > 2. the email-checker and signed-off-by-checker CI *must* pass on that PR > > This still allows you to merge early (i.e., before other CI completes). > That's a different issue, and is probably ok the way that it is currently > handled (i.e., individual developer's discretion -- usually let all the CI > finish, but merge early when the situation warrants it).
I personally have no objections > > -- > Jeff Squyres > [email protected] > > > > _______________________________________________ > devel mailing list > [email protected] > https://lists.open-mpi.org/mailman/listinfo/devel _______________________________________________ devel mailing list [email protected] https://lists.open-mpi.org/mailman/listinfo/devel
