On Tue, Oct 16, 2012 at 12:18 PM, Mark A. Hershberger mah@everybody.org wrote:
On 10/16/2012 11:45 AM, Tyler Romeo wrote:
There should be some general idea of at least what is planned for a release before the code is actually written.
This would mean getting any non-WMF contributors (the volunteers) to spec out what they planned to work on before hand and be committed to actually delivering it.
I'm not sure that is realistic.
It is realistic is getting a schedule for WMF-sponsored work, but a good deal of that is not going to interest the average MW admin since it is focused on Wikipedia.
As a sort of compromise, maybe we could write up a list of new features MediaWiki administrators would find useful a month before the release is planned. By that time, we've got a very good idea of what is going to be in it.
It could be that I'm just too pessimistic, but I think that comes from my introduction to the term "Cookie-Licking".
Indeed, I agree on all the points here. Lots of things happen in development because somebody has been working on something and then commits it. This is perfectly ok--we don't want to discourage anyone by saying "That's not on the plan."
I think the idea of starting the general release notes maybe a month (or two) out from release is a good idea. It allows the release to start taking shape and we can start targeting a sane branch point. It also would happen when we generally start to "slush" master and ask people to hold off on earth-shattering changes since a branch point is coming up.
-Chad