Bharath Rupireddy <bharath.rupireddyforpostg...@gmail.com> writes:
> On Wed, Jan 10, 2024 at 12:54 PM Tom Lane <t...@sss.pgh.pa.us> wrote:
>> Yeah. I'm not quite sure what's a good way to make this work, but
>> it seems like having "make check-world" always invoke it would not
>> be desirable. Making that conditional on an environment variable
>> setting could be a better idea, perhaps?
> It's easy to miss setting the environment variable and eventually end
> up with code incompatible with pgindent committed.
Well, we expect committers to know what they're doing. I'm not
quite suggesting that committers add "export I_AM_A_PG_COMMITTER=1"
in their ~/.profile and then have the Makefiles check that to decide
what tests are invoked by "make check-world" ... but it doesn't
seem like a totally untenable idea, either.
> IMO, running the
> pgindent in at least one of the CI systems if not all (either as part
> task SyanityCheck or task Linux - Debian Bullseye - Autoconf) help
> catches things early on in CF bot runs itself. This saves committers
> time but at the cost of free run-time that cirrus-ci provides.
But that puts the burden of pgindent-cleanliness onto initial patch
submitters, which I think is the wrong thing for reasons mentioned
upthread. We want to enforce this at commit into the master repo, but
I fear enforcing it earlier will drive novice contributors away for
no very good reason.
regards, tom lane
