Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 22 Oct 1998 09:20:14 -0700 (PDT)
From:      Archie Cobbs <archie@whistle.com>
To:        eischen@vigrid.com (Daniel Eischen)
Cc:        current@FreeBSD.ORG, lists@tar.com
Subject:   Re: Another Serious libc_r problem
Message-ID:  <199810221620.JAA01628@bubba.whistle.com>
In-Reply-To: <199810220106.VAA07017@pcnet1.pcnet.com> from Daniel Eischen at "Oct 21, 98 09:06:18 pm"

next in thread | previous in thread | raw e-mail | index | archive | help
Daniel Eischen writes:
> > > > I would argue that for any case that POSIX says results in "undefined
> > > > behavior", and the pthread code can easily detect this case, FreeBSD
> > > > should immediately abort(3). Threads programmers will thank you
> > > > when their bugs are revealed for them.
> > > 
> > > If it's like pthread_mutex_lock(), POSIX will say that pthread_cond_wait
> > > should return EINVAL if it doesn't own the mutex *and* this condition
> > > is detected by the implementation.  Much as we'd like to say "Bad
> > > programmer, Bad!" I don't think POSIX will allow us to with anything
> > > other than an EINVAL return value.
> >
> > What you've described looks like *defined* behavior to me...
> 
> Well, that's what the POSIX spec says.  If you are going to
> detect the condition, then you must return EINVAL.  If you
> are not going to detect the condition, then "undefined
> behavior" occurs.

So, then it's kindof like an honor system :-) who's to know when
you've "detected" it?

Anyway, I conceed. Modify the original idea with the phrase "when
not in conflict with the spec"...

-Archie

___________________________________________________________________________
Archie Cobbs   *   Whistle Communications, Inc.  *   http://www.whistle.com

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-current" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199810221620.JAA01628>