On Wed, Oct 17, 2012 at 03:14:14PM +0100, Jason McIntyre wrote:
> On Wed, Oct 17, 2012 at 09:06:23AM +0200, Otto Moerbeek wrote:
> >
> > Op 16 okt. 2012 om 22:56 heeft Jason McIntyre <j...@kerhand.co.uk> het
> > volgende geschreven:
> >
> > > On Sat, Oct 06, 2012 at 09:40:23PM +0200, Dawe wrote:
> > >> I think this information is already given in the sentences before.
> > >>
> > >> Index: mmap.2
> > >> ===================================================================
> > >> RCS file: /cvs/src/lib/libc/sys/mmap.2,v
> > >> retrieving revision 1.39
> > >> diff -u -p -u -p -r1.39 mmap.2
> > >> --- mmap.2 12 Apr 2012 12:53:27 -0000 1.39
> > >> +++ mmap.2 6 Oct 2012 19:35:32 -0000
> > >> @@ -188,10 +188,6 @@ The symbol
> > >> .Dv MAP_FAILED
> > >> is defined in the header
> > >> .Ao Pa sys/mman.h Ac .
> > >> -No successful return from
> > >> -.Fn mmap
> > >> -will return the value
> > >> -.Dv MAP_FAILED .
> > >> .Sh ERRORS
> > >> .Fn mmap
> > >> will fail if:
> > >>
> > >
> > > philip guenther replied to this:
> > >
> > > No, it isn't. The previous sentences say that MAP_FAILED
> > > is returned on error; this one says that MAP_FAILED is
> > > *only* returned on error. Contrast this to strtoul(), where
> > > ULONG_MAX is returned on error, but can also be returned
> > > on success, meaning you have to do extra steps (clear errno
> > > before, test it afterwards) to detect failure. This sentence
> > > is the one that makes it clear that that isn't the case
> > > with mmap().
> > >
> > > i have to say, i didn;t get that. after discussing it a bit i've come to
> > > see there is an ambiguity in the sentence being proposed for removal - it
> > > can be read two ways.
> >
> > Wondering which two ways. I can only see one. It must be either lack of
> > coffee or too much exposure to formal logic.
> >
> > That said, I like your diff.
> >
> > -Otto
> >
>
> well, i spent a few days arguing with guenther before i saw it. so i'd
> say it's subtle ;) the sentence in question is this:
>
> No successful return from mmap will return the value MAP_FAILED.
>
> what it's trying to say is that "under no cicumstances will a successful
> mmap return MAP_FAILED". however i (and the original poster, i presume)
> read it as "an unsuccessful return from mmap will return MAP_FAILED".
>
> since the latter had already been said, we supposed it to be
> duplication. i'd have zapped it if guenther hadn;t spotted the true
> intent.
>
> jmc
aha, now that is cleared up I'm even more in favor of your diff.
-Otto
