Bruno Haible via Gnulib discussion list writes:
> I promised:
>> - see how this new warning behaves on the Gnulib source code
>
> The two good news are:
> - The warning is really helpful. With it, I could find all relevant uses of
> '0' quickly. Patch attached and committed.
> - There w
The coreutils test mkdir/p-3.sh currently fails on macOS, AFAICT as a
result of the gnulib change:
e4399abd41 (savewd: port to native MS-Windows, 2024-08-27)
The relevant part of the test is:
mkdir no-access || framework_failure_
p=$(pwd)
(cd no-access && chmod 0 . && mkdir -p "
On 11/14/24 11:51, Bruno Haible wrote:
I would mess with it only when a configure test detects that it expands to a
non-pointer. That is, when a conditional expression
(foo ? "x" : NULL)
sollicits a compilation error. This catches exactly the AIX case.
I don't see how it catches the AIX xl
Hi Paul,
On Thu, Nov 14, 2024 at 07:51:24PM GMT, Bruno Haible wrote:
> Paul Eggert wrote:
> > > Maybe we should override NULL on this platform (where it is defined as 0)?
> >
> > I wouldn't mess with NULL, as too much oddball stuff depends on it.
>
> I would mess with it only when a configure te
Paul Eggert wrote:
> > Maybe we should override NULL on this platform (where it is defined as 0)?
>
> I wouldn't mess with NULL, as too much oddball stuff depends on it.
I would mess with it only when a configure test detects that it expands to a
non-pointer. That is, when a conditional expressio
On 11/14/24 11:26, Collin Funk wrote:
Does AIX support C23?
Gnulib's nullptr module supports nullptr even if the underlying platform
doesn't have it yet.
Am 14. November 2024 um 18:43 schrieb "Paul Eggert":
> That 'daylight' business was a fiasco for other reasons, which I had
> been meaning to clean up but haven't gotten around to. I installed the
> attached Gnulib patch for now; please give it a try.
>
> As the Gnulib manual says, nobody should us
Paul Eggert writes:
> On 11/13/24 23:09, Bruno Haible via Gnulib discussion list wrote:
>> Maybe we should override NULL on this platform (where it is defined as 0)?
>
> I wouldn't mess with NULL, as too much oddball stuff depends on it.
In stddef.h we already replace NULL for NetBSD 5.0, so I d
On 11/13/24 23:09, Bruno Haible via Gnulib discussion list wrote:
Maybe we should override NULL on this platform (where it is defined as 0)?
I wouldn't mess with NULL, as too much oddball stuff depends on it.
Why not use nullptr instead? It's the modern replacement.
Am 14. November 2024 um 18:22 schrieb "Markus Mützel":
> The following might be relevant:
>
> configure:79124: checking spelling of daylight variable
> configure:79153: gcc -o conftest.exe -g -ggdb -O0 -pthread -fopenmp
-fexceptions conftest.c -lpthread -lm >&5
> conftest.c: In function 'main'
That 'daylight' business was a fiasco for other reasons, which I had
been meaning to clean up but haven't gotten around to. I installed the
attached Gnulib patch for now; please give it a try.
As the Gnulib manual says, nobody should use 'daylight'.From 72abb08f4495bf232736f4d13a24bced72a9c327
The following might be relevant:
configure:79124: checking spelling of daylight variable
configure:79153: gcc -o conftest.exe -g -ggdb -O0 -pthread -fopenmp
-fexceptions conftest.c -lpthread -lm >&5
conftest.c: In function 'main':
conftest.c:341:8: error: returning 'int * (*)(void)' from a fun
Hello,
I tried to build Octave for Windows after updating gnulib from
6213c5bd72d15ca5e1ea9c34122899e02fed448c to
f3d96537413f1fc0913cd290b1eea35386295b1b.
I'm using mingw-w64 at commit cf5c50cce0b4f8610eafd95443d1d3767de386a5.
Some configure tests are failing now that have been passing previou
Hi Bruno,
On Thu, Nov 14, 2024 at 04:14:26PM GMT, Bruno Haible wrote:
> I promised:
> > - see how this new warning behaves on the Gnulib source code
Thanks!
> The two good news are:
> - The warning is really helpful. With it, I could find all relevant uses of
> '0' quickly. Patch attache
Am 14. November 2024 um 15:31 schrieb "Bruno Haible"
>
> Markus Mützel wrote:
> > If the unit test passes with it, it might be passing erroneously.
>
> No: Before making the fixes and adding the unit test, I verified that
without
> the fixes, the unit test failed. In other words, the unit test is
I promised:
> - see how this new warning behaves on the Gnulib source code
The two good news are:
- The warning is really helpful. With it, I could find all relevant uses of
'0' quickly. Patch attached and committed.
- There were no false positives.
However, there are warnings that are
Am 14. November 2024 um 13:31 schrieb "Bruno Haible":
> No, this is not needed. In a testdir of the modules fseeko, ftello, fsync,
> on mingw:
> - we have REPLACE_FSEEKO=1, hence fseeko.c gets compiled, and it uses
> the mingw fseeko function,
> - the two new unit tests pass.
> - Apparent
Markus Mützel wrote:
> If the unit test passes with it, it might be passing erroneously.
No: Before making the fixes and adding the unit test, I verified that without
the fixes, the unit test failed. In other words, the unit test is able
to detect the bugs we are interested in.
> > - Apparently
Markus Mützel wrote:
> With the same reasoning as for ftello.c, should this also be changed in
> fseeko.c?
>
> diff --git a/lib/fseeko.c b/lib/fseeko.c
> index 2c3b053a3b..9d003208ba 100644
> --- a/lib/fseeko.c
> +++ b/lib/fseeko.c
> @@ -31,7 +31,10 @@ fseeko (FILE *fp, off_t offset, int whence)
Bruno Haible via Gnulib discussion list writes:
> Hi Alejandro,
>
>> We're planning a path for deprecation of 0 as a null pointer constant,
>> in favor of NULL
>
> +1
+1
> For reference, we discussed the use of nullptr in Gnulib [1][2] in the past.
>
>> As a first step, Joseph proposed adding i
20 matches
Mail list logo
