Re: Should contrib modules install .h files? - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Should contrib modules install .h files?
Date
Msg-id 1836.1533221640@sss.pgh.pa.us
Whole thread Raw
In response to Re: Should contrib modules install .h files?  (Robert Haas <robertmhaas@gmail.com>)
Responses Re: Should contrib modules install .h files?
Re: Should contrib modules install .h files?
List pgsql-hackers
Robert Haas <robertmhaas@gmail.com> writes:
> Yeah, I would have voted -1 if I'd realized that it was close.  Now
> we're in a situation where we have patch not everyone likes not only
> in master (which is OK, because we've got a year to decide whether to
> change anything) but also in v11 (where we have a lot less time).
> That's not so great.

It seems like there were two separate areas of disagreement/questioning,
one being the file layout (whether to add per-extension subdirectories)
and then one about how one would actually use this, ie what would the
-I switch(es) look like and where would they get injected.

My impression is that there was consensus for per-extension
subdirectories, but the usage scenario isn't totally designed yet.
In principle, only the layout question has to be resolved to make
it OK to ship this in v11.  On the other hand, if there's no
very practical way to use the per-extension subdirectory layout,
we might have to rethink that layout.  So I'd be happier about this
if that question were answered promptly.  Failing a satisfactory
answer in the near future, I think we need to revert in v11.

Also, "near future" means "before Monday".  I don't want to ship beta3
with this in place if we end up reverting later, because it'd mean
thrashing packagers' file manifests, which they won't appreciate.
It might be best to revert in v11 for now, and then we can put it back
after beta3 if there's agreement that the questions are satisfactorily
resolved.

            regards, tom lane


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: Problems with plan estimates in postgres_fdw
Next
From: Alvaro Herrera
Date:
Subject: Re: Internal error XX000 with enable_partition_pruning=on, pg 11beta1 on Debian