With all errno values defined in , there is no more need to have
these definitions in the replacement. I've applied this:
2008-09-13 Bruno Haible <[EMAIL PROTECTED]>
* lib/sys_socket.in.h (EINPROGRESS, ...): Remove definitions.
*** lib/sys_socket.in.h.orig2008-09-14 04:32:06.0
> With this, the conversion function from a WSAGetLastError() value to an
> errno value is
>e -> (e < 10025 ? e - 1 :
> e == WSAENAMETOOLONG ? ENAMETOOLONG :
> e == WSAENOTEMPTY ; ENOTEMPTY :
> e);
That's still not enough. Looking at [1], I think it should also m
Now, with the module 'errno' providing all POSIX errno values, the module
'EOVERFLOW' is obsolete. I'm removing it:
2008-09-13 Bruno Haible <[EMAIL PROTECTED]>
* modules/EOVERFLOW: Remove file.
* m4/eoverflow.m4: Remove file.
* modules/EOVERFLOW-tests: Remove file.
Bruno Haible <[EMAIL PROTECTED]> wrote:
>> I discovered that Solaris 11's /bin/sh ...
>
> What is Solaris 11? Do you have any references? Neither Wikipedia [1] nor
> the Sun documentation [2] mention such a release.
Hi Bruno,
Opensolaris calls itself SunOS 5.11.
Hi Paolo and others,
Here's the module 'errno'. It provides a complete POSIX .
With this, the conversion function from a WSAGetLastError() value to an
errno value is
e -> (e < 10025 ? e - 1 :
e == WSAENAMETOOLONG ? ENAMETOOLONG :
e == WSAENOTEMPTY ; ENOTEMPTY :
e
Hi Jim,
> I discovered that Solaris 11's /bin/sh ...
What is Solaris 11? Do you have any references? Neither Wikipedia [1] nor
the Sun documentation [2] mention such a release.
Bruno
[1] http://en.wikipedia.org/wiki/Solaris_(operating_system)
[2] http://docs.sun.com/app/docs/prod/solaris
Hi Paul,
I discovered that Solaris 11's /bin/sh exhibits the following
surprising behavior:
$ /bin/sh -c 'umask 22; (umask 0); umask'
That would cause test failures in coreutils, because gl_POSIX_SHELL
accepted /bin/sh. The following change is enough to make
gl_POSIX_SHELL reject t
Eric Blake wrote:
> Tom has (graciously) allowed me access to his Irix 5.3 system to attempt
> to address this.
Thanks for debugging this!
> > The test ends up calling sprintf(tmp, "%Ld", -0.0L), but
>
> excuse the typo; I meant "%Lf", not "%Ld"
>
> > the system sprintf does not know how to pri
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Eric Blake on 9/13/2008 7:11 AM:
> The test ends up calling sprintf(tmp, "%Ld", -0.0L), but
excuse the typo; I meant "%Lf", not "%Ld"
> the system sprintf does not know how to print -0, so the result is
> 0.00 and lacks -. Does it m
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Tom G. Christensen on 8/9/2008 6:41 AM:
>
> On Irix 6.2 I also see problems in gnulib:
> test-vasprintf-posix.c:1338: assertion failed
> /bin/ksh[10]: 14207 Abort(coredump)
> FAIL: test-vasprintf-posix
>
> The vasprintf-posix problem is
Bruno Haible wrote:
> Hi Paolo,
>
> Paolo Bonzini wrote:
>> except the need to
>> move parts of the sys_socket module to errno/stdio/string.
>
> I can do that for you. It's the usual gnulib macrology and substology.
>
>> 2) regarding string, strerror does not need to be replaced under mingw;
>
Hi Paolo,
Paolo Bonzini wrote:
> except the need to
> move parts of the sys_socket module to errno/stdio/string.
I can do that for you. It's the usual gnulib macrology and substology.
> 2) regarding string, strerror does not need to be replaced under mingw;
Huh? strerror is declared by mingw's
Yoann Vandoorselaere wrote:
> Thanks for the quick review, updated path attached!
I made the following modifications:
- Use WINDOWS32 macro instead of USE_WIN32. This minimizes the diffs with
glibc, for easier sync in the future.
- Use alloca() instead of a fixed-size buffer.
- Change to
Yoann Vandoorselaere replied
> > About the second hunk, the problem is more general: not only on mingw,
> > but on all platforms which lack ETIMEDOUT, one needs a substitute. It
> > can be a random positive number. I propose this patch instead.
Not applied. Errno values are better defined in or ,
Bruno Haible wrote:
> Paolo Bonzini wrote:
>> Ok, will do so (s/HAVE_IO_H/__MSVCRT__).
>
> But __MSVCRT__ is specific to mingw: the MSVC compiler does not define it.
>
> OK to commit this patch?
I am not sure if __MSVCRT__ should be defined by the compiler, or rather
by the library (in which cas
Paolo Bonzini wrote:
> ReadFile and WriteFile work under win32 with sockets (learning that is
> what prompted me to submit this finally).
Terrific!! With this, we can completely emulate the POSIX socket APIs.
Bruno
Paolo Bonzini wrote:
> Ok, will do so (s/HAVE_IO_H/__MSVCRT__).
But __MSVCRT__ is specific to mingw: the MSVC compiler does not define it.
(See also http://predef.sourceforge.net/) Some people use wrapper scripts
that make MSVC usable for autoconf based projects (cccl, cc-msvc.sh or similar).
Bett
The attached patch finally provides the native implementation of poll
for Win32.
Testing under native Windows is definitely needed for this, because Wine
has a big tendency to crash on the test:
1) test_tty fails under Wine (so I don't actually know if it works, but
the code was taken from GNU Sm
This patch fixes all the issues that Bruno found, except the need to
move parts of the sys_socket module to errno/stdio/string. I didn't do
that for these reasons:
1) regarding errno, the #defines were already there before the patch;
2) regarding string, strerror does not need to be replaced und
Eric Blake wrote:
> It looks like you failed to check .m4 tests. I'm applying the patch
> below.
Yes, I missed to change the .m4 tests. Thank you for the fix.
> Meanwhile, I have a question about tests/test-signbit.c - your
> patch changed most occurrences of -0.0L to -zerol, to work around one
20 matches
Mail list logo
