Thread: pgsql: Mention aix cc_r is not supported, and why

Log Message:
-----------
Mention aix cc_r is not supported, and why

Modified Files:
--------------
    pgsql:
        configure (r1.413 -> r1.414)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/configure.diff?r1=1.413&r2=1.414)
    pgsql/config:
        acx_pthread.m4 (r1.9 -> r1.10)
        (http://developer.postgresql.org/cvsweb.cgi/pgsql/config/acx_pthread.m4.diff?r1=1.9&r2=1.10)

Am Dienstag, 14. Dezember 2004 13:58 schrieb Bruce Momjian:
> Mention aix cc_r is not supported, and why

I think this change is inappropriate.  First, cc_r is no more or less common
than cc; you have to purchase either one.  Second, cc_r is supported.  We
have one port report that uses it.  I say we just leave it alone and wait for
people to fix the thread support.

--
Peter Eisentraut
http://developer.postgresql.org/~petere/

Peter Eisentraut wrote:
> Am Dienstag, 14. Dezember 2004 13:58 schrieb Bruce Momjian:
> > Mention aix cc_r is not supported, and why
>
> I think this change is inappropriate.  First, cc_r is no more or less common
> than cc; you have to purchase either one.  Second, cc_r is supported.  We
> have one port report that uses it.  I say we just leave it alone and wait for
> people to fix the thread support.
>

Did you look at the patch?  To support it would require us to use a
special CC setting for all cases where we are linking to a threaded
libpq.  I don't think we want to add that ugliness to our code just to
support it, just like we didn't do a similar change to support threading
on unixware.

If we wanted to support it we could easily do so by using PTHREAD_CC in
all the bin makefiles and other places that link to libpq.

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073

Am Dienstag, 14. Dezember 2004 14:10 schrieb Bruce Momjian:
> Did you look at the patch?  To support it would require us to use a
> special CC setting for all cases where we are linking to a threaded
> libpq.  I don't think we want to add that ugliness to our code just to
> support it, just like we didn't do a similar change to support threading
> on unixware.

Maybe we are just having different notions of "supported".  I think people can
already set CC and CFLAGS themselves to get all-threaded builds.  And that is
supported, I think, because we allow users to set CC and CFLAGS.  If we write
"cc_r is not supported", then we might shut out users where we don't have to.
As I said, one user already reported to have used it.

--
Peter Eisentraut
http://developer.postgresql.org/~petere/

Peter Eisentraut wrote:
> Am Dienstag, 14. Dezember 2004 14:10 schrieb Bruce Momjian:
> > Did you look at the patch?  To support it would require us to use a
> > special CC setting for all cases where we are linking to a threaded
> > libpq.  I don't think we want to add that ugliness to our code just to
> > support it, just like we didn't do a similar change to support threading
> > on unixware.
>
> Maybe we are just having different notions of "supported".  I think people can
> already set CC and CFLAGS themselves to get all-threaded builds.  And that is
> supported, I think, because we allow users to set CC and CFLAGS.  If we write
> "cc_r is not supported", then we might shut out users where we don't have to.
> As I said, one user already reported to have used it.

OK, text updated:

# Supporting cc_r would require a special CC in all places that
# use libpq, and that is ugly, so we don't do it.  Users can still
# define their compiler as cc_r to do thread builds of everything.

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073