Date: Tue, 4 Mar 2003 13:34:02 -0800 From: Kris Kennaway <kris@obsecurity.org> To: Roger Hardiman <roger@telepresence.dmem.strath.ac.uk> Cc: Kris Kennaway <kris@obsecurity.org>, ports@FreeBSD.ORG Subject: Re: HEADS UP: Ports freeze for 4.8 Message-ID: <20030304213402.GA93311@rot13.obsecurity.org> In-Reply-To: <006101c2e275$33665270$1464a8c0@VAIO> References: <20030301233233.GA5152@rot13.obsecurity.org> <006101c2e275$33665270$1464a8c0@VAIO>
next in thread | previous in thread | raw e-mail | index | archive | help
--qMm9M+Fa2AknHoGS
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
On Tue, Mar 04, 2003 at 05:40:50PM -0000, Roger Hardiman wrote:
> It looks to be the way the port dependencies work.
> These ports all require the _source_ to be available for the comms/ixj port,
> but not for the port to be actually installed.
There are several problems here:
1) For a few days bento was building packages in a jail instead of a
chroot. This meant they could not communicate with the network after
the 'fetch' phase, so ports that tried to fetch distfiles during
'extract' would fail. Most of these ports were fixed by the
'fetch-required' patch I recently committed to bsd.port.mk, but as
noted in that commit log ports that have a :target were not being
fetched. I have since backed this out, so it's no longer relevant for
these failures.
2) Something about recursive :target dependencies doesn't work as
expected (and never has, AFAIK). In this case there are two levels of
recursion:
openh323proxy:
BUILD_DEPENDS= ${NONEXISTENT}:${PORTSDIR}/net/openh323:build
openh323:
BUILD_DEPENDS= ${NONEXISTENT}:${PORTSDIR}/comms/ixj:patch
and the second one is not called.
I see the same failure mode on my 4.x box when trying to install
openh323proxy without having openh323 already installed. The
workaround is to not recurse :)
There is another set of problems I noticed here:
.if !exists(/usr/include/sys/telephony.h)
.if !exists(/usr/local/include/sys/telephony.h)
BUILD_DEPENDS= ${NONEXISTENT}:${PORTSDIR}/comms/ixj:patch
.endif
.endif
1) Using exists() to decide to add dependencies is bad, because the
list of dependencies may be generated on a different machine than the
package build runs on (e.g. bento).
2) AFAICT FreeBSD has never shipped /usr/include/sys/telephony.h, so the first test is a NOP
3) /usr/local is hard-coded in the second test and will break if the
user has a different $LOCALBASE.
Kris
--qMm9M+Fa2AknHoGS
Content-Type: application/pgp-signature
Content-Disposition: inline
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (FreeBSD)
iD8DBQE+ZRvJWry0BWjoQKURAsPhAJ4mhWJ1yPYgzN3UYH+BGj8BORIAwwCfespl
L+hDXS5pnD1E7+fqtegdN+U=
=nylk
-----END PGP SIGNATURE-----
--qMm9M+Fa2AknHoGS--
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-ports" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030304213402.GA93311>