Date: Tue, 22 Feb 2005 14:31:12 +0100 From: Dejan Lesjak <dejan.lesjak@ijs.si> To: Dag-Erling =?iso-8859-1?q?Sm=F8rgrav?= <des@des.no> Cc: noackjr@alumni.rice.edu Subject: Re: X.Org 6.8.2 - (most probably) final patch Message-ID: <200502221431.12805.dejan.lesjak@ijs.si> In-Reply-To: <86y8dgannw.fsf@xps.des.no> References: <4219D008.6030107@alumni.rice.edu> <Pine.LNX.4.58.0502221218570.4364@mussel.ucc.gu.uwa.edu.au> <86y8dgannw.fsf@xps.des.no>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tuesday 22 of February 2005 11:34, Dag-Erling Sm=F8rgrav wrote:
> David Adam <zanchey@ucc.gu.uwa.edu.au> writes:
> > 17 minutes ~is~ a long time for extraction (particularly over several
> > ports) but I think it's a problem best solved by split distfiles, not by
> > lower compression.
> >
> > Is there any way around it (turning off cleaning, unbzip2ing and
> > modification of the Makefile/distinfo)?
>
> One possibility is to create a port that installs a patched source
> tree into /usr/X11R6/src/${PORTVERSION}, and have the other ports
> BUILD_DEPEND on it and lndir(1) the source code into their WRKDIR.
> This would have the added advantage of eliminating the current mess
> with random xorg ports nicking patches from eachother.
>
> DES
Excellent idea!
I was pondering on how this would handle the case where a patch would need =
to=20
be added to single port, say xorg-vfbserver... If a patch would go to this=
=20
port then lndir wouldn't help, since it would patch the installed source fi=
le=20
(and on next rebuild fail because of previously applied patch). If a patch=
=20
would go into this source port, then lndir would work, but there needs to b=
e=20
a way for xorg-vfbserver to know that the patch has been applied. I was=20
pondering at first that source port would rather install=20
into /usr/X11R6/src/${PORTVERSION}_${PORTREVISION} and xorg-vfbserver would=
=20
depend on this location. But that would require changing all of the other=20
ports to depend on new location of source. I'm now thinking of two=20
possibilities:
=2D defining proper src location in xorg-source port Makefile.inc or some s=
uch=20
and including this in all xorg- ports
=2D installing in /usr/X11R6/src/${PORTVERSION} but with placeholder files=
=20
like /usr/X11R6/src/${PORTVERSION}/.foopatch-applied and depending on that=
=20
(with proper if(source installed) && !exists(.foopatch-applied) echo=20
"somekindofexplanation")
Anyway, this seems doable so far. I'm still hoping that there will be split=
=20
distfiles for this release since this upgrade is now quite far, and I'm not=
=20
sure whether portmgr will want another test run with such change (the=20
packages contents and installed files should stay the same so this is rathe=
r=20
internal xorg- ports change, but configure parts would probably need to be=
=20
reworked and all...). Nevertheless, if nobody comes up with a scenario in=20
which this way would fail I think this is the way to go :)
I was thinking of doing something to use the same source for all of separat=
e=20
ports through using WRKSRC or somesuch, but never thought of actually=20
installing the source. Nifty trick, thanks :)
Dejan
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200502221431.12805.dejan.lesjak>