[mercury-users] posix stuff

[William Lee Irwin III]

On 15-Mar-2001, William Lee Irwin III <wli at holomorphy.com> wrote:
>> I had a little bit of trouble with the posix select module. It
>> didn't seem to update the file descriptor sets which were meant
>> to hold its results.

On Tue, Apr 03, 2001 at 10:19:53AM +1000, Fergus Henderson wrote:
> Are you sure?  The code looks fine to me (although it could
> be documented better).  I think you have just misunderstood
> how it is supposed to work.

The Mercury code and/or supporting C code is correct as far as I can
tell. I've never had to do this sort of thing for socket code before.
I recently got confirmation of this behavior (for code which does not
use the Mercury runtime) from Tim Stark on alt.sys.pdp10 and so this
appears to actually be a workaround for a bug in either some particular
versions of libc or the Linux kernel. There is probably also a race
involving whatever is clobbering select's argument and the copyback in
the workaround code which is hidden by the quick copyback. If I gain
more insight into what is really happening here I'll cc: this list.


Cheers,
Bill
--------------------------------------------------------------------------
mercury-users mailing list
post:  mercury-users at cs.mu.oz.au
administrative address: owner-mercury-users at cs.mu.oz.au
unsubscribe: Address: mercury-users-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-users-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------