Re: buildfarm + meson

Hi,

On 2023-03-09 14:47:36 -0500, Andrew Dunstan wrote:
> On 2023-03-09 Th 08:28, Andrew Dunstan wrote:
> > At this stage I think I'm prepared to turn this loose on a couple of my
> > buildfarm animals, and if nothing goes awry for the remainder of the
> > month merge the dev/meson branch and push a new release.

Cool!

> > There is still probably a little polishing to do, especially w.r.t. log
> > file artefacts.

> A few things I've found:
>
> . We don't appear to have an equivalent of the headerscheck and
> cpluspluscheck GNUmakefile targets

Yes. I have a pending patch for it, but haven't yet cleaned it up
sufficiently. The way headercheck/cpluspluscheck query information from
Makefile.global is somewhat nasty.

> . I don't know how to build other docs targets (e.g. postgres-US.pdf)

There's an 'alldocs' target, or you can do ninja doc/src/sgml/postgres-US.pdf

> . There appears to be some mismatch in database names (e.g.
> regression_dblink vs contrib_regression_dblink). That's going to cause some
> issues with the module that adjusts things for cross version upgrade.

I guess we can try to do something about that, but the make situation is
overly complicated. I don't really want to emulate having randomly differing
database names just because a test is in contrib/ rather than src/.

Greetings,

Andres Freund