Bug 149661 - emerge of java packages fail if placed between ant-core and ant-tasks update
|
Bug#:
149661
|
Product: Gentoo Linux
|
Version: 2006.0
|
Platform: All
|
|
OS/Version: Linux
|
Status: RESOLVED
|
Severity: major
|
Priority: P2
|
|
Resolution: FIXED
|
Assigned To: java@gentoo.org
|
Reported By: trenton.d.adams@gmail.com
|
|
Component: Development
|
|
|
URL:
|
|
Summary: emerge of java packages fail if placed between ant-core and ant-tasks update
|
|
Keywords:
|
|
Status Whiteboard:
|
|
Opened: 2006-09-30 14:46 0000
|
I've made an attachment because I've got a lot of output.
I think this could make a problem when gen-2 goes stable and people start
upgrading their packages... Not sure what to do about it, maybe have new
ant-core block old ant-tasks? Yeah blockers confuse people and they go filling
bugs and ask on IRC/forums but there's already a blocker for java-config vs
java-config-wrapper in the upgrade path, so one more wouldn't hurt maybe?
(In reply to comment #3)
> I think this could make a problem when gen-2 goes stable and people start
> upgrading their packages... Not sure what to do about it, maybe have new
> ant-core block old ant-tasks?
People are filing invalid bugs about mismatched -core and -tasks all the time,
blockers are still better than leaving this broken. Alternatively, ebuilds
should depend just on dev-java/ant metabuild instead of depending directly on
ant-core/ant-tasks?
I don't think the bug reports are invalid unless users have keyword
inconsistencies, like ant-core ~arch and ant-tasks or ant arch. But even with
good keywords you can run into problems if portage issues something between
ant-core and ant-tasks update (which can also cause upgrade/downgrade loops and
other fun stuff). Making everything depend on ant is against the idea of having
it splitted (because ant-tasks have many and some even fetch restricted deps
etc, and you can't build those deps without having ant-core first). Soon even
ant-tasks will get splitted, but don't worry, I'm doing it so it won't try to
load mismatched versions. For the time being, blocker sounds best... Nichoj?
Yes, we should go with the blocker until we have the totally modular ant-tasks.
The block "!<dev-java/ant-tasks-${PV}" was already present in
ant-core-1.6.5-r2, I've added it to -r14 too.