On 04/17/2011 12:15 PM, Bruno Haible wrote:
> Hi Eric,
>
> There were indeed some mistakes in the test. Now I've debugged it and added a
> test for non-blocking sockets as well.
Thanks again for tackling all of this.
>
> As is, the socket test fails on Cygwin 1.5.x. After putting STDOUT_FILENO
Hi Eric,
There were indeed some mistakes in the test. Now I've debugged it and added a
test for non-blocking sockets as well.
As is, the socket test fails on Cygwin 1.5.x. After putting STDOUT_FILENO in
non-blocking mode, get_nonblocking_flag reports that it is in non-blocking mode,
but it still
On 04/10/2011 08:12 PM, Bruno Haible wrote:
> Hi Eric,
>
> I was a bit suspicious whether non-blocking I/O on Woe32 sockets really work.
> So I set out to write a unit test for it. Sockets are a bit more complicated
> than pipes, so I started with the pipes.
>
> Find at
When a package uses the gnulib modules 'pwrite' and 'nonblocking', it ought
to get a pwrite() function that works on non-blocking file descriptors. This
implements it.
2011-04-13 Bruno Haible
Support non-blocking pipe I/O and SIGPIPE in pwrite().
* modules/pwrite (Depends-on):
Hi Eric,
I was a bit suspicious whether non-blocking I/O on Woe32 sockets really work.
So I set out to write a unit test for it. Sockets are a bit more complicated
than pipes, so I started with the pipes.
Find attached this unit test - for testing, not yet ready to be committed -.
Can you
Hi Eric,
Thanks for taking up and improving on my proposal.
> > - use select() or poll() followed by read() or write() if the
> > descriptor
> >is ready,
> > - call read() or write() in separate threads. */
>
> - use aio_*, although that is not as widely portable
Yes, good p
On 03/30/2011 07:46 PM, Eric Blake wrote:
I'm not sure whether block and character
special devices can usefully be set non-blocking
I believe at least /dev/random supports it:
#include
#include
#include
#include
int main()
{
char buf[256];
int fd = open ("/dev/random", O_RDONLY | O_NON
On 03/30/2011 11:46 AM, Eric Blake wrote:
> On 03/29/2011 09:16 AM, Bruno Haible wrote:
>> Paolo Bonzini wrote:
>>> Without guessing what your bias is, I also :) prefer to implement
>>> {g,s}et_nonblock_flag functions. It would use either
>>> SetNamedPipeHandleState or ioctlsocket (using the soc
> in sockets.c to detect sockets, and then GetFileType to detect pipes if
>> it fails).
>
> Here's proposed code to that effect.
Let's get it checked in.
> Note that the getter function cannot
> be implemented: How to determine whether a Woe32 socket is non-blockin
with WSAEINVAL.
>>
>> To set the socket back to blocking mode, an application must first
>> disable WSAEventSelect by calling WSAEventSelect with the
>> lNetworkEvents parameter equal to zero.
>>
>> Using this trick will allow to ask the socket (but not thread safe :S)
&
is trick will allow to ask the socket (but not thread safe :S)
I am stupid It will not work
Bastien
>
> Bastien
>
>>
>>
>> nonblocking.h
>>
>> /* Non-blocking I/O for pipe or socket descriptors.
>> Copyright (C) 2011 Free Softw
e WSAEventSelect by calling WSAEventSelect with the
lNetworkEvents parameter equal to zero.
Using this trick will allow to ask the socket (but not thread safe :S)
Bastien
>
>
> nonblocking.h
>
> /* Non-blocking I
ect pipes if
> it fails).
Here's proposed code to that effect. Note that the getter function cannot
be implemented: How to determine whether a Woe32 socket is non-blocking?
nonblocking.h ========
/* Non-blocking I/O for pipe o
Eric Blake wrote:
> Mingw has named pipes, but they appear to
> reside in a different namespace than the normal file system, and the
> gnulib mkfifo() implementation current always fails with ENOSYS on
> mingw. Unless we find a way to expose named pipes in the normal file
> system on mingw, then y
On 03/29/2011 06:45 AM, Eric Blake wrote:
> Mingw supports named pipes (witness the mkfifo gnulib function),
Oops, hit send too soon. Mingw has named pipes, but they appear to
reside in a different namespace than the normal file system, and the
gnulib mkfifo() implementation current always fails
vertheless in gnulib we try to offer the POSIX APIs, if possible.
>
> There are three APIs for non-blocking I/O on Unix:
> 1) fcntl F_SETFL O_NONBLOCK (POSIX, SysV)
> 2) fcntl F_SETFL O_NDELAY (older SysV)
> 3) ioctl FIONBIO (BSD 4.2)
4) SOCK_NONBLOCKL
Bastien
> According t
t) or ioctl FIONBIO.
>
> So, barring bugs that we find during unit testing, we need
> - a replacement for fcntl F_SETFL O_NONBLOCK that supports mingw,
> - extend lib/open.c so that it handles O_NONBLOCK on mingw (presumably
> by failing, because Win32 does not support non-
rts mingw,
- extend lib/open.c so that it handles O_NONBLOCK on mingw (presumably
by failing, because Win32 does not support non-blocking I/O on
regular files).
I agree that get/set_nonblocking functions are a good step towards this.
Bruno
--
In memoriam Rachel Levy <http://en.wikipe
18 matches
Mail list logo
