[PATCH v2 1/2] timezone: sync to tzdb 2026b

[Adhemerval Zanella Netto]

On 19/05/26 10:29, Yury Khrustalev wrote:
> On Tue, May 19, 2026 at 10:17:14AM -0300, Adhemerval Zanella Netto wrote:
>> ...
>>>>>> +static void *
>>>>>> +mempcpy(void *restrict s1, void const *restrict s2, size_t n)
>>>>>> +{
>>>>>> +  char *p = memcpy(s1, s2, n);
>>>>>> +  return p + n;
>>>>>> +}
>>>>>> +#endif
>>>>
>>>> This breaks the build for Hurd (error: static declaration of 'mempcpy' 
>>>> follows non-static declaration).
>>>
>>> I bumped into this as well. It blocks testing of my patches. It would be
>>> nice if patches were build-tested for most supported targets using the
>>> build-many-glibcs script before merging into master. I think this is 2nd
>>> time this happens over recent couple of weeks. Perhaps we need to add
>>> build-many-glibcs to the CI?
>>
>> Joseph already runs https://sourceware.org/pipermail/libc-testresults/, adding
>> build-many-glibcs on CL/CI could be too expensive.
> 
> Yes, but this is not pre-commit. Master is broken too often. So, either
> authors should check locally or we need CI. Checks of Hurd take a few
> minutes to run.
> 

I did check for Hurd, although with a pre-built toolchain.  It can become
quite unfeasible to check on all variations and gcc versions on each patch;
I think for general development the current CI is suffice (where is build
with a system toolchain).