Larry McVoy writes:
:-)With regard to the lockd stuff.  SGI, at least that part of SGI that I
:-)have some imfluence over, is moving towards a model where we implement 
:-)an idea and give away our implementation as part of a standardization
:-)effort.

It would be really, really good if any implementation we had was compatible
with Sun's, not least because we have a sparc port (and no SGI port) for
NetBSD and Sun's are so common in network environments. But not only that,
it would be nice to have a bit of extra functionality so that nasties like
close one fd unlocks all locks on that file didn't happen (or is that so
ingrained into people's coding that it couldn't be changed). There's a
program I'm currently developing on a Sun where we have to go through all
sorts of hoops because locking is per process not per fd.

:-)Lockd has always been a royal pain in the but.  Some of the best people
:-)at Sun fucked it up several times; it wasn't because they were stupid, it
:-)was because it is a hard problem.

Yes, we've had our share of problems here. We even have our own special
patch which never seemed to become public (it's not in 4.1.4, although it
is finally fixed as of 5.3 or 5.4 - don't remember which).

Callum Gibson                                             callum@bain.oz.au
Fixed Income Division, DB Bain & Co.                          61 2 258 1620
## The opinions presented herein do not represent those of DB Bain & Co ###