Hi Bruno,
I pushed the patch. I'll let you decided if you want to force a CI run
or wait until the scheduled run.
Bruno Haible writes:
> There are two known ways to build on Android [1].
>
> Maybe there is also a third one, with cross-compilation from an x86_64 system?
> I don't know.
Thanks f
Collin Funk wrote:
> I don't know how to test on Android
There are two known ways to build on Android [1].
Maybe there is also a third one, with cross-compilation from an x86_64 system?
I don't know.
Bruno
[1] https://lists.gnu.org/archive/html/bug-gnulib/2022-12/msg00100.html
Hi Collin,
Thanks for working on this.
> How does the attached patch look for fixing Windows? I haven't pushed
> yet.
Looks good. Please push. Then we'll see whether the CI reports that we
have overlooked something.
Bruno
Hi Bruno,
Bruno Haible writes:
> Since it is easy to forget, let me update the doc regarding all these
> winsock functions.
Thanks.
How does the attached patch look for fixing Windows? I haven't pushed
yet.
On Windows link to -lws2_32 to fix that build error. I've tested it
using mingw and wi
Collin Funk wrote:
> I forgot that you have to link to ws2_32 even for these
> functions on Windows. I'll work on fixing that today. Thanks!
Since it is easy to forget, let me update the doc regarding all these
winsock functions.
2024-08-11 Bruno Haible
doc: Update regarding Windows
Hi Bruno,
Bruno Haible writes:
>For a function, the float-argument test and the evaluates-arg-only-once
>test naturally succeed.
>
>I believe the situation is good here because these 4 functions have been
>heavily used over the last 30 years, and problems like evaluates-arg-
>
PS: I attached the logs for mingw. But the situation on MSVC is the same.
Collin Funk wrote:
> Doing a separate for each seems excessive.
I agree. If a package needs one of these functions, the odds are high that
it needs some of the other three functions as well.
The second patch, with the tests, is fine.
However, the first patch is not sufficient:
1) The portabilit
