On Feb 10, 2014 10:07 AM, "Colin Walters" <walters@verbum.org> wrote:
>
> On Tue, Feb 4, 2014 at 8:54 AM, Josh Boyer <jwboyer@fedoraproject.org> wrote:
>>
>> So, we really kind of need to settle on something and get started.
>
>
> I find the idea that the long and historic relationship between GNOME and Fedora could turn around so quickly like that to be very strange - it's a bit like having a friend who suddenly meets someone else they find interesting, and maybe you get a polite wave when you're passing.
>
> You'd be left wondering what you did wrong...

I think (hope?) I addressed this part in my reply to Richard.

> Further, I really dislike the mindset where it's all about switching between pre-formed but completely different things.  There is a whole spectrum of options in between, such as small forking. 
>
> I think this is something where technology drives culture - packages *punish* forking - you have to tediously rename all of the upstream source code so that the files don't stomp on each other, just for the completely obscure use case of having multiple desktops "installed" at the same time.
>
> Which then in turn makes it *much, much harder* to merge back.  Instead, packages reward writing completely new implementations.   Of course, I come at this from the OSTree perspective, which makes it pretty easy to have scalable branches that you can switch between, instead of requiring co-installation.
>
> And finally, the relative omission of the fact that Red Hat Enterprise Linux 7 has *already forked* in this discussion is kind of odd, to say the least.  Personally I think classic mode is a visible symbol of the malfunctioning feedback mechanism.  Or really, not even feedback - it should be about cooperation, with actual *code* flowing both ways.

I agree that the small forking possibility is extremely important here.  This is not a "pick an upstream and ship it" effort.  We are looking for a basis to start from.  The ability to modify to best meet our end goals is key and that will include adaptations and small diversions in whatever upstream is used.

Your comments around packaging and ostree are certainly valid as well, but I would prefer to tackle one problem at a time :).

josh