Date: Mon, 31 Dec 2001 13:38:53 -0800 (PST) From: Julian Elischer <julian@elischer.org> To: John Baldwin <jhb@FreeBSD.org> Cc: Poul-Henning Kamp <phk@critter.freebsd.dk>, cvs-all@FreeBSD.org, cvs-committers@FreeBSD.org, Julian Elischer <julian@FreeBSD.org>, Greg Lehey <grog@FreeBSD.org> Subject: Re: cvs commit: src/share/man/man3 queue.3 Message-ID: <Pine.BSF.4.21.0112311328270.94344-100000@InterJet.elischer.org> In-Reply-To: <XFMail.011231123912.jhb@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
There is code depending on this all over the place (not just in FreeBSD)
but it's not documented..
better to document it than to just allow it to be a guess.
On Mon, 31 Dec 2001, John Baldwin wrote:
>
> On 30-Dec-01 Poul-Henning Kamp wrote:
> > In message <20011230094936.K90696@monorchid.lemis.com>, Greg Lehey writes:
> >>On Saturday, 29 December 2001 at 8:55:05 +0100, Poul-Henning Kamp wrote:
> >>> In message <200112282200.fBSM0Rw61760@freefall.freebsd.org>, Julian
> >>> Elischer wr
> >>> ites:
> >>>> julian 2001/12/28 14:00:27 PST
> >>>>
> >>>> Modified files:
> >>>> share/man/man3 queue.3
> >>>> Log:
> >>>> Document some behaviour I'm depending on.
> >>>> (TAILQ_FOREACH leaves the variable NULL if there were no more to do)
> >>>
> >>> I don't like this.
> >>
> >>Do you have a reason for not liking it?
> >
> > Yes, I don't like statements with weird sideeffects. If julian
> > wants the sideeffect, he should not use a _FOREACH() macro to hide
> > the internals of the loop.
Most looping constructs in C and in FreeBSD have a correct completion
method.
It would be more surprising to me if it DIDN'T have one than if it did.
In this case, I dispute that this is just a side effect.
It is an exit condition.
> >
> > _FOREACH() is for the plain & simple traversal of a queue(3) stucture,
> > anything more should explicitly state intent.
>
> I would agree except for code like this:
>
> TAILQ_FOREACH(p, &list, p_next)
> if (some_test_on(p))
> break;
> if (p != NULL) {
> /* Found one, now do something. */
> }
This is the case I'm using it for.
except I'm doing:
TAILQ_FOREACH(p, &list, p_next)
if (some_test_on(p))
break;
if (p != NULL) {
/* Found it, now do something. */
}
lots of code that MAY clear p
if (p == NULL) {
/* didn't find it, or it was disqualified */
}
>
> You could work around this via a goto instead I suppose:
>
> TAILQ_FOREACH(p, &list, p_next)
> if (some_test_on(p))
> goto found;
> goto not_found
> found:
> /* Do something */
> not_found:
>
> But that's uglier IMO.
Definitly.
It is ALWAYS possible to do it with more complicated constructs,
but "why?".
>
> --
>
> John Baldwin <jhb@FreeBSD.org> <>< http://www.FreeBSD.org/~jhb/
> "Power Users Use the Power to Serve!" - http://www.FreeBSD.org/
>
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0112311328270.94344-100000>