Peter Jeremy wrote:
On 2009-Apr-13 12:35:05 +0100, Matthew Seaman <[email protected]> 
wrote:
The patch Edwin applied before broke under QAT testing because the
www/rt38 port layout is now completely different to www/rt36 and
earlier, breaking the dependency checking. www/rt36 is on the old
layout, so should be fine.  Even so, internal changes in rt38 code
mean that the various graphs this port would have produced no longer
work correctly, so best to leave it depending on www/rt36 instead.

Is it worth explicitly making it conflict with www/rt38?  At present,
if you already have rt38 installed, p5-RTx-Statistics will build and
then fail to install or work with a non-obvious error message.  A
CONFLICTS line could explicitly explain what the problem is.


Yes.  That would be nice. Unfortunately it won't work: you'ld still see exactly
the same output -- the conflicts message between rt36 and rt38 when rt36 is
attempted to be installed as a BUILD_DEPENDS of p5-RTx-Statistics.  It's a 
rather
a shame that CONFLICTS is not applied until just before the install phase for 
any
port -- sometimes that's right and proper, other times it results in a wasted
build of a port that will refuse to be installed.

        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

Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to