You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm working on a fork of these additions and I'm actively using them in my computer lab. I'm hoping to slowly build on those changes/improvements. However, I'm pretty much a noob and can only do basic git operations at the moment. Is there a way for me to pull request a fork of a fork into the oupstream master? Or do you have any other suggestions how I could contribute? The next major thing I'm going to work on is the web interface, because I have a lab with multiple students trying to use netrender simultaneously and the web interface doesn't facilitate that very well. Another fear about pull request is that I'm just a hack and don't really know what I'm doing =P
@kroko thanks! I'm not familiar with the blender dev process. Let me know if you need testers and such. I'd be more than glad to help in any way I can.
@tylerecouture I have a strong feeling that instead of upstreaming we should write new one from scratch. I remember my feelings when I went through the code after forking the project. Existing plugin has many issues and needs much work
base core logic has some flaws;
UX, the buttons are laid out unlogically, mish/mash;
same GUI button in different situations / app state will yield different action which is bad practice / user unfriendly and I know such action that can be stopped only by closing Blender;
GUI needs state checks in many situations (related to previous);
more user friendly and fast(!) way to switch between working and just pushing job (which also means syncing even more stuff from workspace render current params);
if I remember correctly my changes might have broken headless netrender instantation (I just ignored them altogether), but it is actually a no go in the future, at least for upstream;
total responsive web interface rewrite;
here are so many small little annoying things that can and should be fixed, i.e., if you try to netrender project with textures, netrender should warn and supply a button to bundle them into project before submitting to master, as well as bigger ones - path inconsistency, easier tile rendering, doing simulations calculations in netrender a.o.
IMHO the best way is to leave as it is. I'd strongly suggest waiting for 2.8. to build upon the much-(new)user-fiendlier paradigm. Eevee is close to, but NOT 100% realtime, so it might also benefit from netrendering on superhard scenes, especially if result is composited. 2.8 will also have changes in Python API. Then, based on the new world building up a new plugin from scratch is IMHO the correct way. It is not like this project has 100k lines of code :) Also, this project is not in vain as we can take the best from it and learn from the worst, and build upon that. Plus it does not mean "new plugin" as in git repo/name/location - it can be the same repo, just different branch, that when done and tested can be made master and then suggested for the upstream.
Activity
musm commentedon Jun 24, 2017
cc @tylerecouture
tylerecouture commentedon Jun 24, 2017
I'm working on a fork of these additions and I'm actively using them in my computer lab. I'm hoping to slowly build on those changes/improvements. However, I'm pretty much a noob and can only do basic git operations at the moment. Is there a way for me to pull request a fork of a fork into the oupstream master? Or do you have any other suggestions how I could contribute? The next major thing I'm going to work on is the web interface, because I have a lab with multiple students trying to use netrender simultaneously and the web interface doesn't facilitate that very well. Another fear about pull request is that I'm just a hack and don't really know what I'm doing =P
kroko commentedon Jul 6, 2017
@musm i needed these features fast back then and did not care about upstreaming (netrender is official addon - Blender Addons: included in official releases on blender.org, strictly reviewed). you are right, it should be upstreamed (reference https://wiki.blender.org/index.php/Dev:Doc/Process/Addons#Repositories), but then again IMHO it is time for overhaul. if you check gitlogs https://developer.blender.org/diffusion/BA/history/master/netrender/ then nothing serious has been done on it since 2012.
@tylerecouture if this is pushed upstream, then total hack may not be welcomed by blender devs 😉
musm commentedon Jul 6, 2017
@kroko thanks! I'm not familiar with the blender dev process. Let me know if you need testers and such. I'd be more than glad to help in any way I can.
tylerecouture commentedon Aug 2, 2017
@kroko, do you mind if I try to get your additions pushed upstream?
kroko commentedon Apr 4, 2018
@tylerecouture I have a strong feeling that instead of upstreaming we should write new one from scratch. I remember my feelings when I went through the code after forking the project. Existing plugin has many issues and needs much work
IMHO the best way is to leave as it is. I'd strongly suggest waiting for 2.8. to build upon the much-(new)user-fiendlier paradigm. Eevee is close to, but NOT 100% realtime, so it might also benefit from netrendering on superhard scenes, especially if result is composited. 2.8 will also have changes in Python API. Then, based on the new world building up a new plugin from scratch is IMHO the correct way. It is not like this project has 100k lines of code :) Also, this project is not in vain as we can take the best from it and learn from the worst, and build upon that. Plus it does not mean "new plugin" as in git repo/name/location - it can be the same repo, just different branch, that when done and tested can be made master and then suggested for the upstream.
tylerecouture commentedon Apr 4, 2018
Thanks for the reply! I'm about halfway through a webinterface re-write but got distracted a couple months ago.... I'll focus on finishing that.