Win32 status

Nick C will be delighted to hear that Win32 now passes 1/2 of socketpair
tests. The infamous $! test #14 fails but should be fixable.

Changed tests to "skip" the SOCK_DGRAM tests - not clear that winsock
does UDP so cannot emulate it ...

(
  how does one have more than one SKIP: {} block in a test ?
  is the SKIP: label special ?
)

Big snag now is that thread cv_stress test now hangs (for at least 4 hours)
and ^C kills whole harness so don't know about ones beyond that.

-- 
Nick Ing-Simmons
http://www.ni-s.u-net.com/

0
nick
1/12/2002 1:42:07 PM
perl.perl5.porters 47378 articles. 0 followers. Follow

1 Replies
160 Views

Similar Articles

[PageSpeed] 11

On Sat, Jan 12, 2002 at 01:42:07PM +0000, Nick Ing-Simmons wrote:
> 
> Nick C will be delighted to hear that Win32 now passes 1/2 of socketpair
> tests. The infamous $! test #14 fails but should be fixable.

Hurrah. And Schwern has made Test::More distinguish between undef and '',
and do == tests. Now all I have to do is find a sensible job and my happiness
will be complete (1)

> Changed tests to "skip" the SOCK_DGRAM tests - not clear that winsock
> does UDP so cannot emulate it ...

Merijn found that the original hello world program did compile and
print hello world twice on Cygwin, so that suggests that windows does.

grep on the whole cygwin source code for DGRAM finds this

winsup/cygwin/include/cygwin/socket.h:#define SOCK_DGRAM        2              /* datagram (conn.less) socket   */
winsup/w32api/include/lmwksta.h:#define WKSTA_NUMDGRAMBUF_PARMNUM 31
winsup/w32api/include/winsock.h:#define SOCK_DGRAM      2
winsup/w32api/include/winsock2.h:#define SOCK_DGRAM     2

(I have not attempted to compile Cygwin source code on a non-x86, non Windows
machine)

Nicholas Clark

1: Allegedly. There may well be a couple of other things.
-- 
ENOJOB http://www.ccl4.org/~nick/CV.html
0
nick
1/13/2002 3:48:32 PM
Reply: