Date: Mon, 14 Jun 2010 18:15:58 +0000 From: Alexey Dokuchaev <danfe@FreeBSD.org> To: Anton Berezin <tobez@FreeBSD.org>, "Philip M. Gollucci" <pgollucci@p6m7g8.com>, ports-committers@FreeBSD.org, cvs-ports@FreeBSD.org, cvs-all@FreeBSD.org, portmgr@FreeBSD.org Subject: Re: cvs commit: ports/devel/p5-DateTime-Format-Excel Makefile distinfo Message-ID: <20100614181558.GA76481@FreeBSD.org> In-Reply-To: <20100614174430.GA82322@heechee.tobez.org> References: <201006141204.o5EC4q5J002813@repoman.freebsd.org> <4C166555.4020004@p6m7g8.com> <20100614174430.GA82322@heechee.tobez.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Jun 14, 2010 at 07:44:30PM +0200, Anton Berezin wrote:
> On Mon, Jun 14, 2010 at 05:22:29PM +0000, Philip M. Gollucci wrote:
> > On 06/14/10 12:04, Anton Berezin wrote:
> > > | +RUN_DEPENDS= ${BUILD_DEPENDS}
> > I really thought we were going in the other direction of not setting a
> > depends = to another depends?
>
> This particular commit did not change the status quo. The change was done
> to silence portlint.
>
> As for the substance of setting the two to be the same, I am afraid I missed
> the relevant discussion - the one which reached the consensus that it is not
> a good idea. Could you point me towards it?
It is not a good idea because of make(1) being lazy when evaluating
assignments, which in turn can result in rdeps being polluted by bdeps
when the latter is being altered somewhere deep in bpm. The right thing
to do here is probably not to avoid assigning rdeps to bdeps completely,
but to do it correctly with immediate expansion thereof:
RUN_DEPENDS:= ${BUILD_DEPENDS}
This is being very common lack of knowledge for lots of our fellow
committers, to my regret. :-(
Didn't we have portlint(1) check or PHB paragraph about it?
./danfe
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20100614181558.GA76481>