The current release and backport policy follows the idea of "The old tree's dead wood is the young plants' humus.", where the dead old tree is D6 core, and the young plants are D6 contrib and D7/D8 core. And soon D7 will be the dead tree. Dead, because it does not receive any improvements except for bugfixes.
There are good reasons for this strategy: Dead means stable and reliable. A project in active evolution is like a moving target for contrib developers, a frozen project is something you can build a house on. And secondly, with every new feature there is a chance to introduce bugs and break stuff for existing sites - so, better stick with what you have.
The downside is, that we are stuck with non-optimal solutions for years:
- A not up-to-date jquery
- Plain unsalted and unstretched md5 password hashing (only few site owners will bother to install the safe login module)
- etc
I wonder if Drupal can at some day get to a point where more than just bugfixes can be backported into the active major release. The only rules would be "stay compatible with contrib" and "implement in the new version first, then backport". Extending the API would be allowed, just not breaking it.
Maybe Drupal needs some other things before this can be a realistic vision - better testing framework, more encapsulation, strict API... but I think it is a vision worth to consider.
Comments
Comment #1
donquixote commentedTo clarify:
There can still be a feature freeze for the time before the public release of a new version, to allow intensive testing. The question is what comes after that.
Comment #2
marvil07 commentedComment #3
damien tournoud commentedClosing old discussion.
Comment #4
donquixote commentedI dunno.
Just because this is old, does not mean it is fixed.
The problem does still exist, though this day there are probably more mature or more interesting ideas around, about how to approach this.
http://groups.drupal.org/node/178314
Probably gdo is a better place for this discussion.
Ideal thing for me would be to make this a duplicate of something else.
Otherwise, "wontfix" would be the correct status.
The final tradeoff could be that we don't do anything at all, but we should allow the discussion to happen.