Peter Eisentraut <peter@eisentraut.org> writes:
> I don't have a good sense of what you are trying to optimize for. If
> it's the mainline build-on-every-commit type, then I wonder how many
> commits would really be affected by this. Like, how many commits touch
> only a README file. If it's for things like the cfbot, then I think the
> time-triggered builds would be more frequent than new patch versions, so
> I don't know if these kinds of optimizations would affect anything.
As a quick cross-check, I searched our commit log to see how many
README-only commits there were so far this year. I found 11 since
January. (Several were triggered by the latest round of pgindent
code and process changes, so maybe this is more than typical.)
Not sure what that tells us about the value of changing the CI
logic, but it does seem like it could be worth the one-liner
change needed to teach buildfarm animals to ignore READMEs.
- trigger_exclude => qr[^doc/|\.po$],
+ trigger_exclude => qr[^doc/|/README$|\.po$],
regards, tom lane