GNU bug report logs - #78659
[PATCH] Pacify Clang's -Wformat-signedness.

Previous Next

To reply to this bug, email your comments to 78659 AT debbugs.gnu.org.
There is no need to reopen the bug first.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Collin Funk <collin.funk1 <at> gmail.com>
To: bug-gnu-emacs <at> gnu.org
Subject: [PATCH] Pacify Clang's -Wformat-signedness.
Date: Sat, 31 May 2025 22:16:15 -0700

[Message part 1 (text/plain, inline)]

Hi,

When compiling with './autogen.sh && ./configure CC=clang && make':

    [...]
    xfns.c:2041:6: warning: format specifies type 'unsigned int' but the argument has type 'int' [-Wformat]
     2040 |           sprintf (css, "scrollbar slider { background-color: #%02x%02x%02x; }",
          |                                                                ~~~~
          |                                                                %02x
     2041 |                    color.red >> 8, color.green >> 8, color.blue >> 8);
    [...]

This warning is harmless, but interestingly 'gcc -Wformat-signedness'
does not warn about it, but 'clang -Wformat-signedness' does.

Since the color fields of struct XColor are 'unsigned short' and the
constant is an 'int', the expression should be an 'int'.

Anyways, the attached patch silences the warnings.

Collin

[0001-Pacify-Clang-s-Wformat-signedness.patch (text/x-patch, attachment)]

Message #8 received at 78659 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Collin Funk <collin.funk1 <at> gmail.com>,
 Paul Eggert <eggert <at> cs.ucla.edu>
Cc: 78659 <at> debbugs.gnu.org
Subject: Re: bug#78659: [PATCH] Pacify Clang's -Wformat-signedness.
Date: Sun, 01 Jun 2025 09:24:07 +0300

> From: Collin Funk <collin.funk1 <at> gmail.com>
> Date: Sat, 31 May 2025 22:16:15 -0700
> 
> When compiling with './autogen.sh && ./configure CC=clang && make':
> 
>     [...]
>     xfns.c:2041:6: warning: format specifies type 'unsigned int' but the argument has type 'int' [-Wformat]
>      2040 |           sprintf (css, "scrollbar slider { background-color: #%02x%02x%02x; }",
>           |                                                                ~~~~
>           |                                                                %02x
>      2041 |                    color.red >> 8, color.green >> 8, color.blue >> 8);
>     [...]
> 
> This warning is harmless, but interestingly 'gcc -Wformat-signedness'
> does not warn about it, but 'clang -Wformat-signedness' does.
> 
> Since the color fields of struct XColor are 'unsigned short' and the
> constant is an 'int', the expression should be an 'int'.
> 
> Anyways, the attached patch silences the warnings.

Thanks.

Paul, any comments or suggestions?

Message #11 received at 78659 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: Eli Zaretskii <eliz <at> gnu.org>, Collin Funk <collin.funk1 <at> gmail.com>
Cc: 78659 <at> debbugs.gnu.org
Subject: Re: bug#78659: [PATCH] Pacify Clang's -Wformat-signedness.
Date: Sat, 31 May 2025 23:47:15 -0700

On 2025-05-31 23:24, Eli Zaretskii wrote:
> Paul, any comments or suggestions?

As a general rule of thumb, I wouldn't bother pacifying anything but the 
latest GCC, especially to work around false alarms (which these are). 
Since this is isn't the latest GCC and these are false alarms, I suggest 
ignoring the false alarms, or silencing them by using ./configure 
--disable-gcc-warnings.

Come to think of it, when building from Git maybe 'configure' should 
make --enable-gcc-warnings the default only when using recent GCC. 
Currently it also defaults to --enable-gcc-warnings with Clang or with 
older GCC, but that's probably more trouble than it's worth.

Message #14 received at 78659 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Paul Eggert <eggert <at> cs.ucla.edu>
Cc: 78659 <at> debbugs.gnu.org, collin.funk1 <at> gmail.com
Subject: Re: bug#78659: [PATCH] Pacify Clang's -Wformat-signedness.
Date: Sun, 01 Jun 2025 10:29:33 +0300

> Date: Sat, 31 May 2025 23:47:15 -0700
> Cc: 78659 <at> debbugs.gnu.org
> From: Paul Eggert <eggert <at> cs.ucla.edu>
> 
> On 2025-05-31 23:24, Eli Zaretskii wrote:
> > Paul, any comments or suggestions?
> 
> As a general rule of thumb, I wouldn't bother pacifying anything but the 
> latest GCC, especially to work around false alarms (which these are). 
> Since this is isn't the latest GCC and these are false alarms, I suggest 
> ignoring the false alarms, or silencing them by using ./configure 
> --disable-gcc-warnings.

Agreed.

> Come to think of it, when building from Git maybe 'configure' should 
> make --enable-gcc-warnings the default only when using recent GCC. 

I tend to agree, but isn't the latest GCC prone to bugs in this area?

And how can we know what is "recent GCC"?

Message #17 received at 78659 <at> debbugs.gnu.org (full text, mbox):

From: Collin Funk <collin.funk1 <at> gmail.com>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: Paul Eggert <eggert <at> cs.ucla.edu>, 78659 <at> debbugs.gnu.org
Subject: Re: bug#78659: [PATCH] Pacify Clang's -Wformat-signedness.
Date: Sun, 01 Jun 2025 18:04:32 -0700

[Message part 1 (text/plain, inline)]

Hi Eli and Paul,

Eli Zaretskii <eliz <at> gnu.org> writes:

>> Date: Sat, 31 May 2025 23:47:15 -0700
>> Cc: 78659 <at> debbugs.gnu.org
>> From: Paul Eggert <eggert <at> cs.ucla.edu>
>> 
>> On 2025-05-31 23:24, Eli Zaretskii wrote:
>> > Paul, any comments or suggestions?
>> 
>> As a general rule of thumb, I wouldn't bother pacifying anything but the 
>> latest GCC, especially to work around false alarms (which these are). 
>> Since this is isn't the latest GCC and these are false alarms, I suggest 
>> ignoring the false alarms, or silencing them by using ./configure 
>> --disable-gcc-warnings.
>
> Agreed.

How about just disabling this warning if Clang is being used like we do
for many others?

Simple 1 line patch attached.

Collin

[v2-0001-Pacify-Clang-s-Wformat-signedness.patch (text/x-patch, attachment)]

Message #20 received at 78659 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 78659 <at> debbugs.gnu.org, collin.funk1 <at> gmail.com
Subject: Re: bug#78659: [PATCH] Pacify Clang's -Wformat-signedness.
Date: Mon, 2 Jun 2025 00:26:30 -0700

On 2025-06-01 00:29, Eli Zaretskii wrote:
> I tend to agree, but isn't the latest GCC prone to bugs in this area?

Not that I know of. I usually build with it.

> And how can we know what is "recent GCC"?

I suggest worrying only about the most recent stable GCC release. 
Anything older than that is generally not worth worrying about, though 
there may be some exceptions in extremely annoying circumstances. After 
all, these warnings are meant only for developers who should be using 
recent GCC anyway.

Message #25 received at 78659-done <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: Collin Funk <collin.funk1 <at> gmail.com>, Eli Zaretskii <eliz <at> gnu.org>
Cc: 78659-done <at> debbugs.gnu.org
Subject: Re: bug#78659: [PATCH] Pacify Clang's -Wformat-signedness.
Date: Mon, 2 Jun 2025 00:27:31 -0700

On 2025-06-01 18:04, Collin Funk wrote:
> How about just disabling this warning if Clang is being used like we do
> for many others?

Thanks, I installed that. Closing the bug report.

Message #28 received at 78659 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Paul Eggert <eggert <at> cs.ucla.edu>
Cc: 78659 <at> debbugs.gnu.org, collin.funk1 <at> gmail.com
Subject: Re: bug#78659: [PATCH] Pacify Clang's -Wformat-signedness.
Date: Mon, 02 Jun 2025 10:52:14 +0300

> Date: Mon, 2 Jun 2025 00:26:30 -0700
> Cc: collin.funk1 <at> gmail.com, 78659 <at> debbugs.gnu.org
> From: Paul Eggert <eggert <at> cs.ucla.edu>
> 
> On 2025-06-01 00:29, Eli Zaretskii wrote:
> > I tend to agree, but isn't the latest GCC prone to bugs in this area?
> 
> Not that I know of. I usually build with it.
> 
> > And how can we know what is "recent GCC"?
> 
> I suggest worrying only about the most recent stable GCC release. 
> Anything older than that is generally not worth worrying about, though 
> there may be some exceptions in extremely annoying circumstances. After 
> all, these warnings are meant only for developers who should be using 
> recent GCC anyway.

OK, but the question of how to recognize that version still stand.
Assuming we want to enable the warning automatically by some autoconf
test, that is.

Message #31 received at 78659 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 78659 <at> debbugs.gnu.org, collin.funk1 <at> gmail.com
Subject: Re: bug#78659: [PATCH] Pacify Clang's -Wformat-signedness.
Date: Mon, 2 Jun 2025 11:24:57 -0700

On 2025-06-02 00:52, Eli Zaretskii wrote:
> the question of how to recognize that version still stand.

That version recognition is done by configure.ac; see its use of 
gl_GCC_VERSION_IFELSE. Currently the default is to enable warnings (but 
don't fail the build) with any version of Clang, or with GCC 5.3 or 
later. We could easily change the default to some other version.

Message #34 received at 78659 <at> debbugs.gnu.org (full text, mbox):

From: Eli Zaretskii <eliz <at> gnu.org>
To: Paul Eggert <eggert <at> cs.ucla.edu>
Cc: 78659 <at> debbugs.gnu.org, collin.funk1 <at> gmail.com
Subject: Re: bug#78659: [PATCH] Pacify Clang's -Wformat-signedness.
Date: Tue, 03 Jun 2025 14:13:30 +0300

> Date: Mon, 2 Jun 2025 11:24:57 -0700
> Cc: collin.funk1 <at> gmail.com, 78659 <at> debbugs.gnu.org
> From: Paul Eggert <eggert <at> cs.ucla.edu>
> 
> On 2025-06-02 00:52, Eli Zaretskii wrote:
> > the question of how to recognize that version still stand.
> 
> That version recognition is done by configure.ac; see its use of 
> gl_GCC_VERSION_IFELSE. Currently the default is to enable warnings (but 
> don't fail the build) with any version of Clang, or with GCC 5.3 or 
> later. We could easily change the default to some other version.

I see.  Should we perhaps have some procedure in place to update the
cutoff version?  Like in make-tarball.txt, perhaps (since releasing a
new version might be a good time to consider updating this)?

Message #37 received at 78659 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: Eli Zaretskii <eliz <at> gnu.org>
Cc: 78659 <at> debbugs.gnu.org, collin.funk1 <at> gmail.com
Subject: Re: bug#78659: [PATCH] Pacify Clang's -Wformat-signedness.
Date: Tue, 3 Jun 2025 23:05:37 -0700

On 2025-06-03 04:13, Eli Zaretskii wrote:
> I see.  Should we perhaps have some procedure in place to update the
> cutoff version?  Like in make-tarball.txt, perhaps

Wouldn't hurt to add it to a checklist somewhere, I suppose. But since 
this a developer thing, not a user thing, we can change the cutoff any 
time; it doesn't need to be associated with a release. It's more likely 
we'd change the cutoff due to a change in Emacs's bleeding-edge code, 
than due to a release.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.