Re: pkg/41492 (lang/python25 wrongly detects files belonging to py25-subversion)

[Jens Rehsack <rehsack%googlemail.com@localhost>]

The following reply was made to PR pkg/41492; it has been noted by GNATS.

From: Jens Rehsack <rehsack%googlemail.com@localhost>
To: Greg Troxel <gdt%ir.bbn.com@localhost>
Cc: recht%NetBSD.org@localhost, pkg-manager%netbsd.org@localhost, 
pkgsrc-bugs%netbsd.org@localhost, 
 gnats-admin%netbsd.org@localhost, obache%NetBSD.org@localhost, 
gnats-bugs%NetBSD.org@localhost
Subject: Re: pkg/41492 (lang/python25 wrongly detects files belonging to 
py25-subversion)
Date: Thu, 28 May 2009 09:27:27 +0000

 Greg Troxel wrote:
 > I think the problem is only with 'make replace', because python has some
 > code to rebuild .py files in site-packages.  When doing make update or
 > bulk builds the depending packages aren't installed.  Given the
 > unsafe_depends notion, I'd say that python should not write files that
 > are not in its planned set of files (as expressed in PLIST).  If it's
 > necessary to rebuild .pyo and .pyc after updating from, e.g., 2.5.5 to
 > 2.5.6, then a) that's pretty scary and b) pkg_rolling-replace or the
 > manual process one should be following instead after make replace should
 > take care of it.
 
 Absolutely true. I got in mind that maybe py-subversion could create
 compiled modules for it's installed .py-files, too.
 
 Jens