Subject: Re: THREAD SAFETY section
To: Chris G. Demetriou <firstname.lastname@example.org>
From: Guenther Grau <Guenther.Grau@bk.bosch.de>
Date: 09/10/1998 20:26:41
Chris G. Demetriou wrote:
> Guenther Grau <Guenther.Grau@bk.bosch.de> writes:
> > Sounds great to me, except that it should be near the top, like
> > on Solaris,
> Uh, on what pages on Solaris is it near the top?
Hmm, I checked Solaris-2.5.1, which is what I've been working
on for the last year or so. I just checked in 2.6 and it's like
you describe it. (Maybe it's time to upgrade to 2.7, to find it
different again? ;-)
> The pages that i've seen (on 2.6) have it in the "ATTRIBUTES" section,
> always near the bottom of the manual page. (In particular, i just
> checked waitpid() and fprintf(), but that's been true for the other
Well, now that's funny :-) I also had a look at fprintf to verify :-)
> Thinking about it, most uses of the manual pages won't want to know
> this information; why should it be at the top?
Hmm, maybe it was, when multi threading was "new". Must have been
by the time when Solaris 2.5 came out ;-) Ok, I'm convinced. (S)he,
who knows that thread safety is important, will find the information
in the man pages, and the others probably don't care. It's just handy
when you are doing thread-development to easily spot the information,
but the others don't need to be bothered with this, I guess.
> FWIW, i'm not convinced that giving each separate attribute we might
> want to describe its own section is a good thing. I do kinda like the
> Solaris way of describing attributes:
> The tables themselves are ugly, and some of the attributes wouldn't
> apply to our code, but I think overall that the idea makes sense.
Yes, just had a look at it, too, and I like this concept as well,
while I also have to agree that the tables look ugly :-)