Re: Frustrating issue with PGXS - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Frustrating issue with PGXS
Date
Msg-id 28888.1182784122@sss.pgh.pa.us
Whole thread Raw
In response to Re: Frustrating issue with PGXS  (Magnus Hagander <magnus@hagander.net>)
Responses Re: Frustrating issue with PGXS
List pgsql-hackers
Magnus Hagander <magnus@hagander.net> writes:
> On Mon, Jun 25, 2007 at 10:43:37AM -0400, Tom Lane wrote:
>> Correct --- how else is it going to find out where the installation is?

> You can specify the full path in the command to pg_config in your Makefile.
> It'd be neat if the makefile could fint that out and use it for further
> references to pg_config. I haven't had time to look into if this is at all
> possible, though.

The trick is to override this bit in Makefile.global:

PG_CONFIG = pg_config

bindir := $(shell $(PG_CONFIG) --bindir)
datadir := $(shell $(PG_CONFIG) --sharedir)
... etc ...

It might be better if the standard invocation of a PGXS build were not

ifdef USE_PGXS
PGXS := $(shell pg_config --pgxs)
include $(PGXS)

but something like

ifdef USE_PGXS
PG_CONFIG := pg_config
PGXS := $(shell $(PG_CONFIG) --pgxs)
include $(PGXS)

to sync these invocations of pg_config with the ones in
Makefile.global.  I'm not sure though how to get this setting to
override the one in Makefile.global ... or should we just remove
that one?
        regards, tom lane


pgsql-hackers by date:

Previous
From: Andrew Dunstan
Date:
Subject: Re: msvc and vista fun
Next
From: Tom Lane
Date:
Subject: Re: Soft deadlocks