Thanks for the comments/questions.  A couple of points in response:
1. About the #if defined (_HP_aCC) || defined (_HP_cc)
We agree.  That line can go.
2. About the TMPFILE definition change from .c to .h:  Unfortunately,  the -V option does not work the same way across HP compilers (it works as expected with HP aCC, but not with HP cc).  

We have a couple of alternatives here.  
a. Using `uname` (to look for HP-UX)  and checking to see if $CPP is set to /usr/bin/cc or /opt/aCC/bin/aCC. Not good.
b. Doing nothing. The TMPFILE mod is a trivial change to make with each new version of PostgreSQL that we build for HP-UX.  (We had in fact suggested this in our original request email). So if  you make no changes to the TMPFILE code, that will be fine too.
ViSolve OpenSource Team
Received: (qmail 20935 invoked from network); 30 Nov 2003 04:33:45 -0000
Received: from (HELO ([])
          (envelope-sender <[EMAIL PROTECTED])
          by (qmail-ldap-1.03) with SMTP
          for <[EMAIL PROTECTED]; 30 Nov 2003 04:33:45 -0000
Received: (from [EMAIL PROTECTED])
by (8.11.6/8.11.6) id hAU4Xcs26486;
Sat, 29 Nov 2003 23:33:38 -0500 (EST)
From: Bruce Momjian <[EMAIL PROTECTED]
Subject: Re: [BUGS] (Modified) Patch request for PostgreSQL 7.4 for HP-UX
To: ViSolve Open Source Team <[EMAIL PROTECTED]
Date: Sat, 29 Nov 2003 23:33:38 -0500 (EST)
CC: PostgreSQL-patches <[EMAIL PROTECTED],
   Peter Eisentraut <[EMAIL PROTECTED]
X-Mailer: ELM [version 2.4ME+ PL108 (25)]
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset=US-ASCII
ViSolve Open Source Team wrote:
Subject: PostgreSQL Patch: Modified Test-and-set routine for HP-UX
(IA-64) specifically for the HP-C compiler
With reference to Tom Lane's response to our previous patch request  (at
this is a modified and more focussed patch request for PostgreSQL for
for HP-UX 11i V2  for the Intel Itanium architecture  (known to the
PostgreSQL code as IA-64).
Peter mentioned:
>What is this line all about?
>+#if defined(__HP_aCC) || defined(__HP_cc)
>There are no other compilers supported, so this seems redundant.
OK, I will remove this check.
Peter also mentioned:
>You're doing this:
I'm afraid this will not fly, because calling the preprocessor is only
portable on .c files.  Generally, it's also unwise to rely in this kind
of subtle side effect.  We need a general solution.
Can you give us a way to test for your preprocessor?  Is there some
unique output like --version or -V that we can test and use *.h only
in that case?
  Bruce Momjian                        |
  [EMAIL PROTECTED]               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073


Reply via email to