Re: pgsql: ecpg: Output dir, source dir, stamp file argument for preproc/*.

Hi,

On 2022-07-18 18:36:50 -0400, Tom Lane wrote:
> I wrote:
> > Andres Freund <andres@anarazel.de> writes:
> >> Grepping for 'no warnings' shows that check_rules.pl has the same
> >> issue. Perhaps worth fixing at the same time, given how closely related they
> >> are?
> 
> > Ah, I didn't think to look around.  I'll see what I can do there.
> 
> Oh, that one's only hiding *one* sloppy spot.  However, while testing
> it I realized that the Makefile rule has two serious deficiencies:
> 
> * check_rules.pl not listed as a prereq, so nothing happens if you
> change it and say "make".
> 
> * check_rules.pl should be run first.  As-is, if it complains,
> preproc.y has already been updated and so another run will succeed.

Good catches.


> You might wanna check your meson conversion for the same bugs.

Don't think it has. check_rules.pl and parse.pl are run by separate build
rules, with a stamp file marking the success of check_rules.pl. However, they
can currently run concurrently, with the preproc.c generation depending on
both.

I guess it'd be a tad cleaner if the dependency instead were to run
check_rules.pl before parse.pl, the degree of parallelism probably doesn't
matter here.

Both rules depend on the scripts.

Greetings,

Andres Freund