keep ppport.h in sync on all branches

From: Andrew Dunstan <andrew(at)dunslane(dot)net>
To: PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org>
Subject: keep ppport.h in sync on all branches
Date: 2010-03-30 12:33:03
Message-ID: 4BB1EF7F.1030003@dunslane.net
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers


Some recent off-list discussion has suggested that we should keep
plperl's copy of ppport.h in sync on all branches, just as we do with
timezone data files. So when we update it on HEAD we should
simultaneously update the back branches. It's a derived file, produced
by the perl module Devel::PPPort, so there is really no maintenance
effort involved. In general I think this is a good idea. The whole idea
of ppport.h is to aid in backwards compatibility.

Here is some info from (my version of) the Devel::PPPort docs:

Perl’s API has changed over time, gaining new features, new
functions, increasing its flexibility, and reducing the impact on
the C namespace environment (reduced pollution). The header file
written by this module, typically ppport.h, attempts to bring some
of the newer Perl API features to older versions of Perl, so that
you can worry less about keeping track of old releases, but users
can still reap the benefit. [...] Currently, Perl versions from
5.003 to 5.10.0 are supported.

(Of course, we don't want to support anything nearly as old as 5.003.
Perl 5.6.0 was released in 2000, so I wouldn't be inclined to look
kindly on complaints about perl releases before that, since it's well
before the release date of the earliest live branch we have. But that
doesn't concern us here.)

That means if we keep this file up to date in all branches, we can be
much less worried about whether backporting a patch, as for example Tom
recently did here:
<http://git.postgresql.org/gitweb?p=postgresql.git;a=commit;h=ef0ec92a57e565a668851f23a894c4b96d60fec8>,
might break a build made with an older version of perl.

As we recently did a significant upgrade of ppport.h on HEAD, I
therefore propose to apply that change to all the live branches.

Another point: as this is a derived file, I think it should be excluded
from pgindent runs.

Comments?

cheers

andrew

Browse pgsql-hackers by date

  From Date Subject
Next Message Robert Haas 2010-03-30 13:23:38 Re: Questions about 9.0 release note
Previous Message Stefan Kaltenbrunner 2010-03-30 12:15:34 Re: Parallel pg_dump for 9.1