http://www.slony.info/bugzilla/show_bug.cgi?id=40

           Summary: libpq thread test insufficient
           Product: Slony-I
           Version: 1.2
          Platform: All
        OS/Version: Linux
            Status: NEW
          Severity: critical
          Priority: medium
         Component: slon
        AssignedTo: slony1-bugs at lists.slony.info
        ReportedBy: peter_e at gmx.net
                CC: slony1-bugs at lists.slony.info
   Estimated Hours: 0.0


config/acx_libpq.m4 is apparently of the opinion that it only needs to verify
that libpq is compiled thread-safely on aix, mingw, and solaris.  I think that
is insufficient.  I have found a posting about this on the mailing list:

"""
Date: Donnerstag, Juni 14, 2007 11:26:04 -0400
From: Andrew Sullivan <ajs at crankycanuck.ca>
To: slony1-general at gborg.postgresql.org
Cc:
Subject: Re: [Slony1-general] idle in transaction on subscriber

On Thu, Jun 14, 2007 at 04:19:20PM +0200, Jacques Caron wrote:
> PostgreSQL library (libpq) was not compiled with
> --enable-thread-safety, and this causes random breakage in slon. The

You're not actually supposed to need to use that switch on platforms
where it's not checked, on the grounds that thread safety is _always_
on there.  What's your platform and compilier?  It may be that we
need more tests.
"""

It is definitely wrong to assume that platforms other than those listed have
thread-safety always enabled.  Linux is a counterexample.

I think it would be most prudent to require thread-safety on all platforms
except those where you have documented evidence that it is not necessary.  I
don't know any among the more common Unix-like platforms that would fit that
description, however.


-- 
Configure bugmail: http://www.slony.info/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.
You are the assignee for the bug.