Hi Eric,
Eric Blake writes:
> On 12/07/2010 02:24 PM, Ludovic Courtès wrote:
>> Hello,
>>
>> Commit c98cb289 changed #ifdefs in unistr.in.h to #ifs. As a
>> side-effect, this prevents compilation of users of this file, such as
>> Guile, with ‘-Wundef’:
>
> Unfortunately, that points to a bug i
* modules/pipe-posix: New file.
* m4/unistd_h.m4 (gl_UNISTD_H_DEFAULTS): Set default.
(gl_UNISTD_H): Check for declaration.
* modules/unistd (Makefile.am): Substitute it.
* lib/unistd.in.h (pipe): Provide it for mingw.
* doc/posix-functions/pipe.texi (pipe): Update documentation.
* MODULES.html.sh
Jim, Bruno, and Paolo,
I want to eventually get to the point where open(O_CLOEXEC) can be
supported in gnulib for use in libraries like LGPLv2+ libvirt. Along
the way, I will want to make open (LGPLv2+) depend on cloexec (currently
LGPLv3+), which in turn is a thin wrapper around fcntl (currently
On 12/10/2010 03:36 PM, Eric Blake wrote:
> Are there any objections to relaxing the license?
None here.
By the way, it bugs me a bit that in the normal
(e.g., Linux or Solaris) case, set_cloexec_flag
issues two system calls when one would do. Could
this be fixed at some point? One advantage o
Eric Blake wrote:
> The module name 'pipe'
> is already claimed for an unrelated application-level wrapper around
> both pipe(2) and the notion of a child process.
Since at this point we are not considering moving either all POSIX
related modules or all application-level modules to a new naming
co
Hi Eric,
> since it is just a #define rather than a replacement function, it
> fails the POSIX aspect of being able to pass a function pointer to
> pipe around
And it also fails the signature test in C++ mode: When I add these lines
*** tests/test-unistd-c++.cc.orig Sat Dec 11 02:47:25 201
Eric Blake wrote:
> testing I/O through the pipe is impossible without
> either threading or spawning a child process, all of which seem rather
> complex given how simple the module itself is.
A partial test is better than none. In fact, we already have a test for the
pipe2() function. It's easy t
Hi Eric,
> cloexec (currently LGPLv3+), fcntl (currently LGPLv3+).
> Are there any objections to relaxing the license?
Fine with me. lib/fcntl.c contains some originally GPLed code from me
(the DuplicateHandle stuff: originally in lib/w32spawn.h under GPL,
moved to lib/cloexec.c on 2009-12-05, b
