MIME-Version: 1.0
References: <760c9dd1-2d80-c223-3f90-609b615f7918@2ndquadrant.com>
 <127d468dd80c7b1b4c616e111431faf60e091697.camel@postgresql.org>
 <d6d249c4-517c-ff99-5935-891960af9ebd@2ndquadrant.com> <TYAPR01MB5866A71B744BE01B3BF71791F5AEA@TYAPR01MB5866.jpnprd01.prod.outlook.com>
 <CA+hUKGKvkPh7XENM8QaF=B5-V84A6bThMLsyv1JM5XMcg9FeKw@mail.gmail.com> <2530095.1733411471@sss.pgh.pa.us>
In-Reply-To: <2530095.1733411471@sss.pgh.pa.us>
From: Thomas Munro <thomas.munro@gmail.com>
Date: Fri, 6 Dec 2024 15:44:20 +1300
Message-ID: <CA+hUKGJLMh+6W5E4M_jSFb43gnrA_-Q6-+Bf3HkBXyGfRFcBsQ@mail.gmail.com>
Subject: Re: MinGW compiler warnings in ecpg tests
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: "Hayato Kuroda (Fujitsu)" <kuroda.hayato@fujitsu.com>, 
	Peter Eisentraut <peter.eisentraut@2ndquadrant.com>, Michael Meskes <meskes@postgresql.org>, 
	"pgsql-hackers@lists.postgresql.org" <pgsql-hackers@lists.postgresql.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Archived-At: <https://www.postgresql.org/message-id/CA%2BhUKGJLMh%2B6W5E4M_jSFb43gnrA_-Q6-%2BBf3HkBXyGfRFcBsQ%40mail.gmail.com>
Precedence: bulk

On Fri, Dec 6, 2024 at 4:13=E2=80=AFAM Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Thomas Munro <thomas.munro@gmail.com> writes:
> > Yeah.  This warning is visible on CI, and on fairywren since its MSYS2
> > upgrade a couple of months ago.  Old MinGW didn't like %lld (I think
> > perhaps the printf from msvcrt.dll from 1996 didn't like it and MinGW
> > knew that), but new MinGW doesn't like %I64d (that's interesting, but
> > not relevant here because %lld is clearly the correct format string,
> > and it works).  We should just revert that change.  Here's a patch.
>
> +1

Thanks for looking.  Pushed, and that fixed that on fairywren.

> > Those were there before the upgrade.  POSIX says that environ should
> > not be declared by a header, but Windows apparently declares it, or at
> > least its cousin _environ, in <stdlib.h> which we include in c.h.  I
> > have no idea why Visual Studio doesn't warn, or why the documentation
> > only tells you about _environ and not environ, or where the macro (?)
> > comes from that renames it, but it passes CI and is
> > warning-free on both toolchains if you just hide the offending
> > declarations.
>
> Isn't this likely to break things for every other Windows toolchain?
> I think the concept might be OK, but we need a tighter #if condition.

Cool, I'll do that for MinGW only then.

> An alternative could be to add the missing dllimport on Windows;
> it's not clear whether other toolchains would care.

I've been digging through its headers (working on a fix for the off_t
bug report) and noticed in passing that it probably thinks we're
re-declaring this function:

https://github.com/mingw-w64/mingw-w64/blob/8bcd5fc1a72c0b6da3147bf21a4a494=
c81d14fae/mingw-w64-headers/crt/stdlib.h#L221

Seems like a good idea to give that a wide berth :-)