Messages in this thread | | | Date | Wed, 29 Jun 2011 00:54:41 +0200 (CEST) | From | Jiri Kosina <> | Subject | Re: [PATCH 03/29] gma500: Ensure the frame buffer has a linear virtual mapping |
| |
On Tue, 28 Jun 2011, Alan Cox wrote:
> > I am still wondering how come that this is causing trouble to anyone > > though -- is anyone really developing real code on top of linux-next > > (which should be there to cross-check merge problems between subsystems > > and test functionality) instead of particular subsystem tree? > > I run -next on a regular basis. I'm actually doing a lot of Linux mid > development on it because I need bits going via various trees from x86 to > staging to input together.
Yes, cross-tree development is definitely a mode in which I can imagine linux-next could be used as a base for actual code development (and probably the only one which is justifiable).
> I'm not btw saying your approach is wrong - in fact I imagine its the > only way to make it managable for some things but in the gma500 case at > least and I suspect much of staging it tends to cause merge problems. It > would be helpful if you route any gma500 bits via me because of the > amount of change in that subtree.
Yeah, thanks.
As said already -- normally I don't accept staging bits at all (for a reason), that was a mistake. Sorry for that.
On Tue, 28 Jun 2011, Alan Cox wrote:
> > I can either drop the gma500 bits I have queued now (and stop applying > > anything touching it, hard rule), or you rebasing on top of Greg's staging > > tree instead of linux-next (and I sorting out the merge conflict later). > > If you can drop the gma500 bits and send them my way I'll merge them via > the gma500 pile.
Sure. I have already pushed out a tree with those reverted and will be sending you the patches separately in a second.
-- Jiri Kosina SUSE Labs
| |