Delaying the Alpha1?

[Sebastian Kügler]

Hi,

I'm pondering how much is the Thursday-to-be-tagged alpha worth, vs being a 
pain in the ass. We're facing a number of problems right now, let me list 
them:

- We need some frameworks in newer-than-alpha1-version, notably kwindowsystem
- plasma-framework needs to be repackaged, anyway
- everybody is testing with latest frameworks, so even if we manage to build 
against the alpha1 of frameworks, how useful is testing then?
- we have a whole bunch of pending work waiting

You see, we've already learned by just trying to do an Alpha, and we're facing 
a number of problems. That's (almost) fine, it's why we do pre-releases in the 
first place. 
I'm pondering the possibility to delay the first alpha until we have a new 
alpha of frameworks, and then pushing it shortly thereafter, reducing the risk 
of a needed delta.

The disadvantage is that we delay wide-spread testing, the upside is that the 
whole stack is much cleaner and actually possible to be sensibly packaged.

This would mean:

- getting all pending work in quickly, getting the split done, all before end 
of this month

- no changes to plasma-framework or any other framework that is required for 
the Alpha between 28 April (tagging of frameworks alpha2) and Plasma Next 
Alpha 1 tagging (could happen the next day, or as soon as frameworks alpha2 
has been smoke-tested. Plasma Next packages have to work on latest frameworks 
pre-release

- paying more attention to changes in frameworks to make sure the above works

- we either skip on alpha, or we squish the schedule to make it fit (schedule 
could use some aligning to frameworks releases, anyway), (I'd prefer the 
latter)

I'm undecided, but have been for two days, so I thought I'd bring it up here. 
What do you think?
-- 
sebas

http://www.kde.org | http://vizZzion.org | GPG Key ID: 9119 0EF9