[GNC-dev] merging failing tests

Geert Janssens geert.gnucash at kobaltwit.be
Thu Sep 6 02:27:07 EDT 2018


Op woensdag 5 september 2018 23:12:31 CEST schreef Carsten Rinke:
> Why not merging a failing test? Why waiting for the code fix?

Aside from John's explanation I would add this as a reason:
Maint and master are integration branches. Our intention is to keep them in a 
releasable state all the time. That is at any point in time we should be able 
to take their current state and release it.

I wrote "intention" because in practice new commits can always introduce bugs 
that are not immediately detected. Especially as our test coverage is far from 
complete. So we are not there yet. But the intention is there and a branch 
with failing tests is not in a releasable state. So failing tests are ok on 
feature branches but they should be fixed (or dropped/disabled if irrelevant 
or not critical) before they can be merged.

Geert




More information about the gnucash-devel mailing list