When to branch off 2.9 (was: Re: 2.8.7 and 2.9 release plan)

[Boudewijn Rempt]

On Mon, 24 Nov 2014, Friedrich W. H. Kossebau wrote:

> Am Montag, 24. November 2014, 00:50:18 schrieb Jaroslaw Staniek:
>> How about keeping the "master always stable" motto and porting in a branch?
>> Cherry picking the results once they are stable?
>
> "Master always stable" is a good point, and I subscribe to that.
>
> But: I would argue that with the port to a new platform (which I consider
> qt5/kf5) there is an exception. Because "master" rules. Any new stuff usually
> has to make sure it does not break stuff in "master", not the other way
> around. But the port is not just a new feature, which simply has to be
> improved until it is regression free. The port is the new "master". And all
> other code has to make sure it integrates with that.
>
> Any commits to master while the port is going on will only complicate things,
> because they have to be integrated into the ported code as well, which will be
> more work due to all the changed code lines in the port.
>
> So to get the port done as quickly and clean as possible, I would vote for a
> complete freeze of master, until the port is done (which would be roughly a
> month I hope). And if master is frozen anyway, the port could also be directly
> done there.
>

Yes, experience from the first Qt5 port really shows that there is no 
other option. You cannot cherry-pick commits that change pretty much every 
other line, as the porting from, for instance, KUrl to QUrl will do. 
Freeze, run the scripts, make build, unfreeze -- as we already decided at 
the Calligra sprint.

Boudewijn