Concerned about the status of ECS/DOTS

The past month, I've been using ECS/DOTS, and though complex, I really love squeezing out high performant code. However, I've noticed none of the DOTS packages on Unity's Github have been updated in over a month. This is coupled with me telling someone a few days ago about Unity's benefits, only to be met by someone tweeting me how Unity Technologies has a tendency to abandon incomplete features. I wasn't given any examples, but that month+ long halt in DOTS/ECS activity seems like a red flag to me.

 

Ah it's that time again for another state of DOTs thread! I'm just joking around. If you spend some time following this forum you'll notice these come along once or twice every month. Going on memory most recent updates have had 1,2 or maybe 3 months between so I wouldn't consider that a red flag really.

Here are some other related threads on this topic...

https://forum.unity.com/threads/state-of-ecs-dots.1044877/#post-6794069

And in the same thread a response from the CTO about DOTS from 25th Jan 2021:
https://forum.unity.com/threads/state-of-ecs-dots.1044877/#post-6761749

https://forum.unity.com/threads/how...iew-experimental-phases.1043443/#post-6752782

https://forum.unity.com/threads/should-i-use-dots.1006916/#post-6531878

TLDR: It seems important and a priority but it's going quite a bit slower than originally predicted.

 

Unity is "reviewing how they share fresh code with the users". Or something like that. At least they said that.

 

Sources can be viewed from the package contents instead (or directly from your IDE).
UT probably realized nobody really needs a github repo with all the stuff, and decided to avoid wasting time on it.

 

I see. It wasn't necessarily interest in the changes, but rather, at this point, if there were any at all.

 

A lot of features were delayed, in favor of working on ISystemBase. 0.17 got delayed because of a nasty bug. 0.18, better ISystemBase support, Enable/Disabled components, faster main-thread data acquisition and better Schedule vs. Run performance with low entity counts are just some of the things we are waiting for and we have to be patient about it.

It's not that they are abandoning anything.

 

Alright, so it's just taking a while. That's a relief.

 

And ~60 Developers are working on Dots FULL time!!

 

And invisible stuff like source generators.

 

Best practices involves putting most things as ScheduleParallel though right, and Run a lot less often? Enabling and disabling would be nice, I guess compared to adding a component to exclude it from system queries, this would do it without the added structural changes that would involve. I'm also excited on the faster low entity counts, as generally games can spawn things and scale up, so the lower levels to be faster is great. Lots of those demos were for large numbers, but you generally don't spawn 100k bullets all the time in game-play haha. But having no limits is great.

As for updates, the package manager has definitely not been updating lately, and a lot less last year then the year before. But I think that's fine, since the updates they do push are either broken and fixed quickly, or just contain more changes then before. And the current packages are very stable at least.

 

This. There is no point in Github if they anyway don't take any PRs in it so not using the power of the open source and the community at all. Especially when the (public) source code is available in the package anyway.