Guys,
I read only part of the discussion but seeing the reasons to either drop
the 1.3 compatibility (no resources, too much effort) or keep it
(versioning requirements), both sides have valuable arguments.
My initial thought was similar to Ralph's only he had better arguments ;-) :
- stop the development of Cocoon 2.1 after the release of 2.1.10
- rename the current branch to Cocoon 2.2 without the 1.3 compatibility
(and maybe other minor changes that are now prevented by the versioning
contract)
- rename the current trunk to Cocoon 3.0
- rename what is currently mentioned as Cocoon 3.0 to Cocoon 4.0 (or
whatever).
This makes the expectations of the various versions far more clear than
the current versioning.
As a sidenote: I think this also makes it easier to write migration
paths from 2.1 to 2.2 to 3.0. And I get the feeling that the proposed
2.2 version is/can be closer to the proposed version 3.0 than in the
current situation, thus making the migration a little less hard.
WDYT?
Bye, Helma
Ralph Goers said the following on 18/12/06 09:26:
earlier. I just took a look at
http://cocoon.apache.org/versioning.html. Removing support for JDK 1.3
on 2.1.x seems to violate that document as all patch levels should be
binary compatible. So it seems the only way to remove support for JDK
1.3 would be to move to a new minor version number.
Again, it would also seem to me that the current 2.2 also doesn't
conform to that document as the change from 2.1 to 2.2 seems to me to
something more than minor given how much the core has changed. A change
from 2.1 to 2.2 would imply that nearly all user written components will
still work perhaps only requiring a recompile. While this may or may
not be true - I certainly haven't tried - it is certain that how an end
user application is integrated with Cocoon has completely changed. I'm
also not sure how many existing sitemaps will require modification. So
I would think trunk should be released as 3.0, if for no other reason
than allow the next 2.1 release to be 2.2.0 without jdk 1.3 support.
