On 27.03.2017 20:20, Owen Taylor wrote:
Hi Armin -
Hello,
Thanks for being interested in Mutter and putting forward this proposal! One thing you need to realize is that this is a really hard development task. Without really working through the details, my guess is that getting GNOME to the point where it can start without an Xwayland server running is 3-6 months of full time work for an experienced GNOME Shell developer - someone who has been working on this code for years.
There are several major challenges: 1. The GTK+ backend challenge. Mutter and GNOME Shell use various GTK+ APIs. They use the X11 backend of GTK+ - which requires an X connection. Using the Wayland backend to GTK+ would make no sense. So you'd either have to remove all usage of GTK+ or add some sort of custom GNOME-private backend to GTK+. 2. The code reorganization challenge. You'll simply need to move a lot of stuff around. 3. Lots of different types of residual X usage within Mutter, GNOME Shell and GNOME more broadly that need to be addressed one by one
I was already told that. Original goals included far more than it could possibly be done by me in the GSoC period. It was trimmed down to just one goal and that is to make libmutter and an example compositor not require XWayland, ie move a lot of code around so it runs only when XWayland is available.
I think that a successful SOC project could be done around this subject, but you need to be pretty realistic about what you can accomplish - you should think of the outputs being: A) Documentation of the challenges B) Documentation of possible approaches to each challenge, what works and what doesn't work C) A successful prototype that shows Mutter starting up.
Yes, this is the idea. I have updated my proposal to reflect this. Blog updates will come weekly, with information about past week challenges and accomplishments.
I would consider having a branch of Mutter code that is ready to be merged very much a stretch goal. I like that you are focusing initially on just Mutter - while there are additional challenges within GNOME Shell and the GNOME Desktop environment that will affect how work on Mutter is done starting with just Mutter definitely keeps things more doable. You'll also need to line up a possible mentor - probably one of Florian, Jonas, Rui, or Carlos Garnacho (or Jasper, but he doesn't have much time for Mutter work these days.)
As outlined in my proposal, Jonas and Carlos Garnacho have volounteered to be my mentors before I wrote the proposal.
- Owen
Thank you for taking your time to reply.
Attachment:
signature.asc
Description: OpenPGP digital signature