pgsql/src/test/regress (GNUmakefile pg_regress.sh)

Started by Peter Eisentrautover 25 years ago3 messagescomitters
Jump to latest
#1Peter Eisentraut
peter_e@gmx.net

Date: Tuesday, November 21, 2000 @ 18:40:28
Author: petere

Update of /home/projects/pgsql/cvsroot/pgsql/src/test/regress
from hub.org:/home/projects/pgsql/tmp/cvs-serv27790/src/test/regress

Modified Files:
GNUmakefile pg_regress.sh

----------------------------- Log Message -----------------------------

Allow for different result files when using GCC versus native compiler.

#2Larry Rosenman
ler@lerctr.org
In reply to: Peter Eisentraut (#1)
Re: pgsql/src/test/regress (GNUmakefile pg_regress.sh)

What if the compiler (native) is xlc (AIX?) ?

* petere@postgresql.org <petere@postgresql.org> [001121 17:41]:

Date: Tuesday, November 21, 2000 @ 18:40:28
Author: petere

Update of /home/projects/pgsql/cvsroot/pgsql/src/test/regress
from hub.org:/home/projects/pgsql/tmp/cvs-serv27790/src/test/regress

Modified Files:
GNUmakefile pg_regress.sh

----------------------------- Log Message -----------------------------

Allow for different result files when using GCC versus native compiler.

--
Larry Rosenman http://www.lerctr.org/~ler
Phone: +1 972-414-9812 (voice) Internet: ler@lerctr.org
US Mail: 1905 Steamboat Springs Drive, Garland, TX 75044-6749

#3Tom Lane
tgl@sss.pgh.pa.us
In reply to: Larry Rosenman (#2)
Re: pgsql/src/test/regress (GNUmakefile pg_regress.sh)

Larry Rosenman <ler@lerctr.org> writes:

What if the compiler (native) is xlc (AIX?) ?

At the moment, we only know of a need to distinguish gcc from
vendor-supplied compilers, so I think Peter's solution of "gcc" or "cc"
is OK until proven inadequate. It's going to be fairly difficult to
come up with a portable method of identifying compilers more closely
than that, anyway. Looking at the pathname would be easily fooled,
and moreover not necessarily good enough: HP ships two considerably
different compilers for HPUX, but calls 'em both cc :-(

Let's try this for now, and worry about better ideas when and if we
need one.

(Eventually I think we're going to have to tweak the geometry tests
to round off the last couple digits, which should eliminate most
of the problem anyway...)

regards, tom lane