Received: from malur.postgresql.org ([217.196.149.56]) by arkaria.postgresql.org with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1mXat4-00039a-UD for pgsql-hackers@arkaria.postgresql.org; Tue, 05 Oct 2021 03:13:02 +0000 Received: from localhost ([127.0.0.1] helo=malur.postgresql.org) by malur.postgresql.org with esmtp (Exim 4.92) (envelope-from ) id 1mXat3-00036V-Rt for pgsql-hackers@arkaria.postgresql.org; Tue, 05 Oct 2021 03:13:01 +0000 Received: from magus.postgresql.org ([2a02:c0:301:0:ffff::29]) by malur.postgresql.org with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1mXat3-00036N-K4 for pgsql-hackers@lists.postgresql.org; Tue, 05 Oct 2021 03:13:01 +0000 Received: from sss.pgh.pa.us ([66.207.139.130]) by magus.postgresql.org with esmtps (TLS1.3:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.92) (envelope-from ) id 1mXasw-000656-Qm for pgsql-hackers@postgresql.org; Tue, 05 Oct 2021 03:13:01 +0000 Received: from sss1.sss.pgh.pa.us (localhost [127.0.0.1]) by sss.pgh.pa.us (8.15.2/8.15.2) with ESMTP id 1953CpXt4028443; Mon, 4 Oct 2021 23:12:51 -0400 From: Tom Lane To: =?utf-8?Q?Dagfinn_Ilmari_Manns=C3=A5ker?= cc: pgsql-hackers@postgresql.org Subject: Re: plperl: update ppport.h and fix configure version check In-reply-to: <87y278s6iq.fsf@wibble.ilmari.org> References: <87y278s6iq.fsf@wibble.ilmari.org> Comments: In-reply-to =?utf-8?Q?Dagfinn_Ilmari_Manns=C3=A5ker?= message dated "Tue, 05 Oct 2021 00:40:45 +0100" MIME-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-ID: <4028441.1633403571.1@sss.pgh.pa.us> Content-Transfer-Encoding: 8bit Date: Mon, 04 Oct 2021 23:12:51 -0400 Message-ID: <4028442.1633403571@sss.pgh.pa.us> List-Id: List-Help: List-Subscribe: List-Post: List-Owner: List-Archive: Archived-At: Precedence: bulk =?utf-8?Q?Dagfinn_Ilmari_Manns=C3=A5ker?= writes: > Andres' recent post¹ about PL/Perl on Windows and linked-to² patch > containing an erroneous version check made me realise that we haven't > updated our copy of ppport.h since 2009. Attached is a patch that does > that, and applies code changes suggested by running it. I've tested > `make check-world` with `--with-perl` on both the oldest (5.8.9) and > newest (5.34.0) perls I have handy. I haven't looked at this patch's details, but I can confirm that it also builds and passes regression on prairiedog's 5.8.3 perl. > I also noticed that PL/Perl itself (via plc_perlboot.pl) requires Perl > 5.8.1, but configure only checks for 5.8 (i.e. 5.8.0). The second patch > updates the latter to match. Hmm ... Perl 5.8.x is old enough that probably it matters to nobody in the real world, but if we're going to mess with this, is 5.8.1 the right cutoff? I wonder about this because I believe prairiedog's perl to be the oldest that we have tested in a good long while, so that we shouldn't assert with any confidence that 5.8.1 would actually work. The last time I surveyed the buildfarm's perl versions, in 2017, these were the only 5.8.x animals: Animal | Surveyed build | Configure's version report castoroides | 2017-07-27 12:03:05 | configure: using perl 5.8.4 protosciurus | 2017-07-27 13:24:42 | configure: using perl 5.8.4 prairiedog | 2017-07-27 22:51:11 | configure: using perl 5.8.6 aholehole | 2017-07-27 19:31:40 | configure: using perl 5.8.8 anole | 2017-07-28 00:27:38 | configure: using perl 5.8.8 arapaima | 2017-07-27 19:30:52 | configure: using perl 5.8.8 gharial | 2017-07-27 20:26:16 | configure: using perl 5.8.8 locust | 2017-07-28 00:13:01 | configure: using perl 5.8.8 narwhal | 2017-03-17 05:00:02 | configure: using perl 5.8.8 gaur | 2017-07-22 21:02:43 | configure: using perl 5.8.9 pademelon | 2017-07-22 23:56:59 | configure: using perl 5.8.9 Notice that here, prairiedog is running 5.8.6, which is Apple's vendor-installed perl on that stone-age version of macOS. Shortly after that, I *downgraded* it to 5.8.3. I do not recall exactly why I chose that precise perl version, but it seems pretty likely that the reason was "I couldn't get anything older to build". In short: (a) we're not testing against anything older than 5.8.3 and (b) it seems quite unlikely that anybody cares about 5.8.x anyway. So if we want to mess with this, maybe we should set the cutoff to 5.8.3 not 5.8.1. regards, tom lane