7.2beta3 on Digital Alpha

Just donwloaded and built 7.2beta3 on Compaq Tru64 (Digital Unix) 4.0 on
Alpha architecture and now all regression tests are passed.

Thanks! Has anyone tested on more recent versions of Tru64 (I'm sure
they will work) and/or with gcc on this platform? We had reports
covering 4.0, 5.0, and two compilers for the previous release...

- Thomas

Thomas Lockhart wrote:

Thanks! Has anyone tested on more recent versions of Tru64 (I'm sure
they will work) and/or with gcc on this platform? We had reports
covering 4.0, 5.0, and two compilers for the previous release...

All regression tests passed also with gcc version 2.95.1 19990816
(release) on alphaev67-dec-osf4.0g

--
Alessio F. Bragadini alessio@albourne.com
APL Financial Services http://village.albourne.com
Nicosia, Cyprus phone: +357-22-755750

"It is more complicated than you think"
-- The Eighth Networking Truth from RFC 1925

All regression tests passed also with gcc version 2.95.1 19990816
(release) on alphaev67-dec-osf4.0g

Thanks. I'll note both compilers...

- Thomas

Just tried on Tru64 5.1 using the Compaq C compiler. Failed the float8,
oidjoin and random tests.

*** ./expected/float8-fp-exception.out  Thu Mar 30 02:46:00 2000
--- ./results/float8.out        Wed Nov 28 19:29:33 2001
***************
*** 237,243 ****

  -- test for over- and underflow 
  INSERT INTO FLOAT8_TBL(f1) VALUES ('10e400');
! ERROR:  Input '10e400' is out of range for float8
  INSERT INTO FLOAT8_TBL(f1) VALUES ('-10e400');
  ERROR:  Input '-10e400' is out of range for float8
  INSERT INTO FLOAT8_TBL(f1) VALUES ('10e-400');
--- 237,243 ----

-- test for over- and underflow
INSERT INTO FLOAT8_TBL(f1) VALUES ('10e400');
! ERROR: Bad float8 input format '10e400'
INSERT INTO FLOAT8_TBL(f1) VALUES ('-10e400');
ERROR: Input '-10e400' is out of range for float8
INSERT INTO FLOAT8_TBL(f1) VALUES ('10e-400');

OIDJOIN had this...
! psql: server closed the connection unexpectedly
! This probably means the server terminated abnormally
! before or while processing the request.

--- ./results/random.out        Wed Nov 28 19:33:03 2001
***************
*** 25,31 ****
    GROUP BY random HAVING count(random) > 1;
   random | count 
  --------+-------
! (0 rows)

  SELECT random FROM RANDOM_TBL
    WHERE random NOT BETWEEN 80 AND 120;
--- 25,32 ----
    GROUP BY random HAVING count(random) > 1;
   random | count 
  --------+-------
!     119 |     2
! (1 row)

SELECT random FROM RANDOM_TBL
WHERE random NOT BETWEEN 80 AND 120;

======================================================================

I can post the whole regression.out file if someone needs it.

Jim

Just donwloaded and built 7.2beta3 on Compaq Tru64 (Digital Unix) 4.0

on

Alpha architecture and now all regression tests are passed.

test=# SELECT version();
version
----------------------------------------------------------------
PostgreSQL 7.2b3 on alphaev67-dec-osf4.0g, compiled by cc -std

--
Alessio F. Bragadini alessio@albourne.com
APL Financial Services http://village.albourne.com
Nicosia, Cyprus phone: +357-22-755750

"It is more complicated than you think"
-- The Eighth Networking Truth from RFC 1925

---------------------------(end of

broadcast)---------------------------

Thomas,

see my replace to the original email. Tru64 5.1 fails float8, oidjoin
and random...

Jim

Just donwloaded and built 7.2beta3 on Compaq Tru64 (Digital Unix)

4.0 on

Alpha architecture and now all regression tests are passed.

Thanks! Has anyone tested on more recent versions of Tru64 (I'm sure
they will work) and/or with gcc on this platform? We had reports
covering 4.0, 5.0, and two compilers for the previous release...

- Thomas

---------------------------(end of

broadcast)---------------------------

Just tried on Tru64 5.1 using the Compaq C compiler. Failed the float8,
oidjoin and random tests.

The float8 is certainly OK.

The random test fails, uh, randomly, should succeed sometimes on your
platform, so is probably OK if you try again.

But the oidjoin test is certainly broken. What optimization are you
compiling with? Does it work better if you turn optimization off??

- Thomas

NetBSD used to fail on these tests too, until they implemented soft-float.

On Wed, 28 Nov 2001, Jim Buttafuoco wrote:

"Jim Buttafuoco" <jim@buttafuoco.net> writes:

OIDJOIN had this...
! psql: server closed the connection unexpectedly
! This probably means the server terminated abnormally
! before or while processing the request.

This is bad. There should be a core dump file --- can you provide
a debugger backtrace from it?

regards, tom lane

will do and report...

I used the default configuration. just did a ./configure && gmake &&
gmake check

Just tried on Tru64 5.1 using the Compaq C compiler. Failed the

float8,

my error, the workstation I was compiling on had max_proc_per_user set
to 64. Changed to 1024 and now just float8 and random fail. It looks
like these errors are ok

Jim

"Jim Buttafuoco" <jim@buttafuoco.net> writes:

my error, the workstation I was compiling on had max_proc_per_user set
to 64.

Ah-hah. We probably ought to document that as a possible source of
failure in the parallel regression tests. I think the present script
will start as many as 20 parallel tests, which means 60 parallel
processes under your userid (a backend, a psql, and probably an owning
shell process for each psql). Add in the postmaster, your own shell,
a couple random make subprocesses, and you got trouble.

Changed to 1024 and now just float8 and random fail. It looks
like these errors are ok

You should find that random fails only, um, randomly. If it fails
every time then there's a problem.

regards, tom lane