Re: CVS commit: pkgsrc/mk/flavor/pkg

[David Holland <dholland-pkgtech%netbsd.org@localhost>]

On Sun, Jun 13, 2010 at 12:44:37PM +0200, Joerg Sonnenberger wrote:
 > The currently discussed situation was artificially created and
 > pre-existing cases differ in so far as "ignore the issue" doesn't
 > work for incremental changes.
 > 
 > The situation will get worse at point where pkgsrc does handle ABI
 > compatiblity correctly.

No, it will get worse only if you break it. Nothing about doing a
complete but incremental rebuild is incorrect, particularly on a
machine that's otherwise idle and therefore not going to trip on
intermediate states. This should remain a supported method.

There is no way in which it constitutes "just closing the eyes and
hoping for the best" -- please stop ranting and explain yourself, and
without pretending that it matters that some packages are temporarily
broken during the rebuild process.

(Whether pkg_rr handles all tricky cases on its own without manual
intervention is an entirely separate question.)

 > Note that step (3) doesn't necessarily have to backout all packages,
 > just offending ones. E.g. glib2 itself doesn't have to be removed for an
 > inplace perl update over major versions to work. 

Do we now have special secret sauce in the dependency tracking to
allow this? Please explain how we're supposed to be able to tell
(other than by manual intervention in the list of packages to
incrementally replace, by someone who has specific detailed knowledge)
which perl dependencies are "major", and which are "minor", and how to
extract this secret sauce from the package database. All I see is
pkg_info -n.

-- 
David A. Holland
dholland%netbsd.org@localhost