gzip has the same tables/code as gnulib's crc module,
so while scoping out the gzip change (to use gnulib's crc), I hit this:
crc.c: In function 'crc32_update_no_xor':
crc.c:77:1: error: function might be candidate for attribute 'pure'
[-Werror=sugge
st-attribute=pure]
77 | crc32_update_no_xor
glibc 2.2 was released more than 23 years ago. Workarounds for glibc 2.1
are therefore not needed any more.
2024-09-09 Bruno Haible
striconv, striconveh: Drop workaround for glibc 2.1.
* lib/striconv.c (str_iconv): Remove glibc-2.1 bug workaround.
* lib/striconveh.c (i
And more of the same kind. Solaris 9's successor, Solaris 10, was released
in 2005. Solaris 9 workarounds therefore can also be dropped.
2024-09-09 Bruno Haible
striconv, striconveh, unicodeio: Drop workaround for glibc 2.1.
* lib/striconv.c (mem_cd_iconv, str_iconv): Remove g
On 2024-09-09 07:06, Bruno Haible wrote:
And more of the same kind. Solaris 9's successor, Solaris 10, was released
in 2005. Solaris 9 workarounds therefore can also be dropped.
Yes, sounds good. The rule I suggest is that if a distro's supplier no
longer supports it, we needn't support it eit
On Mon, Sep 9, 2024 at 4:04 PM Paul Eggert wrote:
>
> On 2024-09-09 07:06, Bruno Haible wrote:
> > And more of the same kind. Solaris 9's successor, Solaris 10, was released
> > in 2005. Solaris 9 workarounds therefore can also be dropped.
>
> Yes, sounds good. The rule I suggest is that if a dist
On 2024-09-09 13:38, Jeffrey Walton wrote:
Apple is notorious for deprecating support quickly in an effort to
drive hardware sales.
Sure, but people who run macOS releases before macOS 12 (the oldest
Apple still supports) are kinda on their own no matter what we do.
Although Gnulib still con
Running some gnulib tests with clang's UBSAN, I get this error in the file
test-string-desc.sh.log:
string-desc-contains.c:39:13: runtime error: null pointer passed as argument
1, which is declared to never be null
/usr/include/string.h:391:33: note: nonnull attribute specified here
SUMMARY
On Mon, Sep 9, 2024 at 8:08 PM Bruno Haible wrote:
>
> Running some gnulib tests with clang's UBSAN, I get this error in the file
> test-string-desc.sh.log:
>
> string-desc-contains.c:39:13: runtime error: null pointer passed as
> argument 1, which is declared to never be null
> /usr/include/
Jeffrey Walton wrote:
> >if (needle._nbytes == 0)
> > return 0;
> > + if (haystack._nbytes == 0)
> > +return -1;
> >void *found =
> > memmem (haystack._data, haystack._nbytes, needle._data,
> > needle._nbytes);
> >if (found != NULL)
>
> Clang is complaining haystack._da
