Dear list, Ahem. Let's try that again with the correct spelling of the list address...
The following slave ports are producing an incorrect 'MASTER_PORT' value:
emulators/linux_base-gentoo-stage1 (linux_base-gentoo-stage1-2006.0_2)
emulators/linux_base-gentoo-stage2 (linux_base-gentoo-stage2-2006.0_1)
emulators/linux_base-gentoo-stage3 (linux_base-gentoo-stage3-2006.0_1)
japanese/kinput2-canna (ja-kinput2-canna-3.1_2)
japanese/kinput2-canna+freewnn (ja-kinput2-canna+freewnn-3.1_2)
japanese/kinput2-canna+freewnn+sj3 (ja-kinput2-canna+sj3+freewnn-3.1_2)
japanese/kinput2-canna+sj3 (ja-kinput2-canna+sj3-3.1_2)
japanese/kinput2-canna+sj3+wnn6 (ja-kinput2-canna+sj3+wnn6-3.1_2)
japanese/kinput2-canna+sj3+wnn7 (ja-kinput2-canna+sj3+wnn7-3.1_2)
japanese/kinput2-canna+wnn6 (ja-kinput2-canna+wnn6-3.1_2)
japanese/kinput2-canna+wnn7 (ja-kinput2-canna+wnn7-3.1_2)
japanese/kinput2-freewnn+sj3 (ja-kinput2-sj3+freewnn-3.1_2)
japanese/kinput2-sj3 (ja-kinput2-sj3-3.1_2)
japanese/kinput2-sj3+wnn6 (ja-kinput2-sj3+wnn6-3.1_2)
japanese/kinput2-sj3+wnn7 (ja-kinput2-sj3+wnn7-3.1_2)
japanese/kinput2-wnn6 (ja-kinput2-wnn6-3.1_2)
japanese/kinput2-wnn7 (ja-kinput2-wnn7-3.1_2)
lang/gfortran (gcc-withfortran-4.1.2_20060721)
math/spooles-mpich (spooles-mpich-2.2_4)
science/mpqc-mpich (mpqc-mpich-2.3.0)
x11-toolkits/fltk-threads (fltk-threads-1.1.6_1)
The symptom is that in each case the MASTER_PORT variable gets set to
a fully qualified path like so:
happy-idiot-talk:...ports/emulators/linux_base-gentoo-stage1:% make -V
MASTER_PORT
/usr/ports/emulators/linux_dist-gentoo-stage1/
Whereas the output is meant to be relative to PORTSDIR, like so:
happy-idiot-talk:...ports/databases/mysql50-client:% make -V MASTER_PORT
databases/mysql50-server
The problem in each case is a trailing slash on the value of MASTERDIR set in
the slave port Makefiles:
happy-idiot-talk:...ports/emulators/linux_base-gentoo-stage1:% grep
^MASTERDIR Makefile
MASTERDIR= ${.CURDIR}/../linux_dist-gentoo-stage1/
It's a pretty trivial thing I know, but I noticed it while working on my
FreeBSD::Portindex module, which relies heavily on the value of MASTER_PORT
to find master/slave relations between ports in the latest release. I've
put in a work-around in my code: I'll leave it to the collected wisdom of the
list to decide if this warrants fixing.
Port maintainers were CC'd in my first attempt at sending this.
Cheers,
Matthew
--
Dr Matthew J Seaman MA, D.Phil. 7 Priory Courtyard
Flat 3
PGP: http://www.infracaninophile.co.uk/pgpkey Ramsgate
Kent, CT11 9PW
signature.asc
Description: OpenPGP digital signature
