Resolved Is DOTS being abandoned?

I was recently informed that in a forum post it was implied DOTS is no further a priority for Unity and is being abandoned. Something that I find hard to digest.

Is that true?
Thank you.

 

Short answer no. It's just on hold until they have the current show-stopper figured out.

 

What does this even mean?

 

Mid-year 2020..."No Show"
Late 2021..."Show Stopper"
So uhm thinkin' of changin' my name t' Mater.
It's like Tamater but without the Ta!

 

My understanding is that they reached a point in the DOTS développement that requires more integration with the unity core engine and/or heavy change to DOTS packages.
It means potentially much more breaking issues and DOTS still being in preview they simply don't want to "waste" ressources in solving/explaining all the issue the integration may bring. They prefer to stop delivering/exposing those changes until they have a version they feel is stable enough for "general" public.

It does not mean Unity abandoned DOTS it just mean they temporarily stopped communicating about their progress on the subject.

I understand this is a confusing/stressfull situation but I for one still trust Unity continues développement of DOTS.

They made the choice to communicate early on their intentions and take feedback from the community instead of doing everything internally and "dropping" a finished product on us.
Those who follow the développement of ECS from the early package know that much have already changed and that in some forum thread certain feedback have been considered but require heavy refactoring (manly thinking about the enable/disable component feature).
The more développement progress the more other packages/aspect come into play (physics/net code/graphics/ audio/...) Maybe they raise new challenges that need additional refactoring to solve in a efficient and elegant way.

 

That's hopeful but the omens don't look good, at least while the community continues to be kept in the dark for such a long and extended period. At this point I wouldn't consider DOTS a safe bet for future projects.

 

Note this is all speculation and since beginning of this year there has been no updates to ECS package and no "official" forum posts. You can conclude what you want from this but to me this is dead until I'm proven otherwise.
I think it makes little sense to dig into Unity's ECS now.
First reason is the uncertainty. It would not hurt to tell us what's going on.
Second reason is, that whenever something new will be released it probably contains major changes so your stuff will not work any more.

When I would see this behavior in an Early Access game I would certainly consider it dead. Not sure what other criteria should apply here.
So my suggestion is to don't hold your breath and spend your time with something different. When there is progress and a new ECS version gets released act surprised and rejoice. But the current situation just leads to frustration when you deal with it. So better to avoid IMO.

 

First of all: Unity announced in may that they work on update and near update will be till the end of this year so why you cry that there no any other updates?

Second: Engine is not a shipped Game and games differ from each other i.e. first announcement of CyberPunk2077 was in 2012.

Third: you talk about ECS not about DOTS

Fourth: we know that unity work on rewrite on to C# SourceGenerators and ECS is still in Experimental phase so of course everything will change it must change in not back compatible way. And Unity already told that adding Enable/Disable component ruin mostly every other ECS package and this is hard step, so we need to prepare to do same hard step after update.

Fifth: We know about some shadow of move forward because ECS samples was updated to run on ECS version 0.20.0 recently.

Sixth: Unity patented ECS memory model!!!

Seventh: On SIGGRAPH 2021 unity presented last HybridRenderer 2.0 that is way faster than current one and 60 times faster than URP on CPU side

Dont looks like ECS is abandoned, isn't it?

Conclusion: If you want to know something about ECS you need to follow info on forum not only complain. Just read forum

 

I'm talking about actual updates. As far as I can ascertain latest ECS release is 0.17 from January. I glance about the sticky threads in DOTS subforum from time to time and could not find any more recent news.

I don't talk about announcements but about released Early Access games. When there is no update for them community is also "worried". When there is no update for a while there should at least be some communication about whats going on. Silence is just irritating and frustrating for "fans" and community.

I suspected op to mean ECS since other parts of DOTS are not on lockdown as far as I'm aware. OP would not be the first one to confuse the terms. Can we get some clarification and maybe a link to what "forum post" you refer?

I guess you refer to my "your stuff will not work any more"? That was not a complaint, just an expectation. And the point is still valid. If Op starts now with ECS he will most likely have to rewrite much of his code. He can decide to do it anyway, but he should be aware.

I think I have read what you mean. And I found the situation quite funny. But this "unofficial information" is nothing to place plans on, do you agree?

How many patent owners don't use their patents themself but just "milk" others which want to use it? Not saying that this is happening here. Just that your "argument" is not neccessarily a lifesign for ECS.

I can't keep an eye on EVERY extern information source. I thought if there is something important to know they would post it here in their own forums. Am I naive because of that?

It looks certainly abandoned. At least to me. They changed from an active communication to complete silence. No reasons, no explanations. People asking detailed questions are just ignored. I don't have that trust and "optimism" you seem to have.

That's exactly what op, me and many others want. Knowing something about the state of Unity ECS. And we read the forum and learn nothing new. Just silence and beeing ignored, Thats exactly the point. Even if I see those "breadcrumbs" you presented I draw different conclusion from them. The longer this silence goes the more trust is lost.

Another point is I don't want to invest too much time in reading about this when future seems so dire and unclear. It would be just a waste of time if it is already cancelled. So please excuse that I'm not up to date on every little information spread across the internet.

I think it's natural that people want to know what is going on. UT refuses this. I'm not aware of any reason for this. I just think they should not treat their customers/users/community this way.

Anyway. Another thread like this got locked already. So this is the last thing I will say about this topic. If UT does not want ECS to look dead they should just talk to us about it. If they don't talk it looks dead and it is "fair" and logical to draw this conclusion. Until then your guess is as good as mine.

 

...

ROFL.
https://forum.unity.com/threads/notice-on-dots-compatibility-with-unity-2021-1.1091800/

You guys take turns to post BS "is DOTS abandoned?" fake worrying posts? What good does it make?

 

First is it not fake worrying, it is a legitimate question. .I do not worry.
I couldn't care less if they discontinue it.
I just need to know for professional reasons if it is worth investing time on it or not.

Second, I have not seen any other similar posts, I would not bother posting otherwise.

Third, thanks for the link. It's helpful. Just lose the attitude. We can all carry one.
Remain professional.

 

Well, I'm sorry, but the "I heard somewhere that it's abandoned." isn't professional either. Instead of coming to the forum, check the pinned threads and notice the one with "notice" and reading it posting the 15th "is it abandoned?" thread. For professional reasons.

 

Do you think this half year old thread offers "reasons" or "explanations"? The only word which could be seen as those (with much fantasy) is "crossroads". Whatever that means in this context.
More important is what MartinGram posted there:

They cannot give the slightest update since then? People asked specific questions and got no response. You may think the "right" moment is just not here yet. I think they have forgotten about this.

What good does this silence make? How is the sudden change in UT's communication scheme good? Which purpose does it serve?
On the other hand "complaining" is feedback. Most people usually state when something bothers them. Why should that not be possible here?
How should UT know what people expect when noone says what he/she wants?
How do you solve problems/issues/misunderstandings with other people when not by talking?

There has been some "enthusiasm" about DOTS-ECS. I just wonder if it was justified. This behavior is just odd and not what I would expect from a company so dependent on their community.

 

I'm just wondering here. If you have an appointment with your MD do you nag him until the date is due? I wonder because when unity said "no updates until we have something, and don't expect anything in this year" you keep "complaining" about where the next update is.

As for the OP original question:
As for every experimental package it's not recommended to be used in production. That said DOTS have its parts. Jobs and Burst are released already so it should be ok to use them, but ECS and derivatives are still in the experimental phase. There are use cases on another thread of using them in production but as far as I know, developers needed some extra time to make it possible.

 

Can I have a link to that post please?

This would help soothe the forum - they're being a bit irate because it's a shock to them to think about DOTS being cancelled.

Totally agree- the responses have been a bit tetchy!

 

Advancing DOTS is the only In Progress item in the 3D Characters and Animation Roadmap. See this post about animation. The NetCode subforum has very timely developer responses to issues and talks about whether or not something has been fixed internally yet or not. Unity is clearly still working on DOTS.

The reality is that DOTS is preview tech and Unity has a lot of work in order to make it usable for anyone on a team that does not have a knowledge of low level programming. I assume this has a been large part of the work that has been on going this year. Unity will release news on DOTS when they are ready to; creating and hijacking forums asking for Unity to release news now does not serve any purpose but to drive speculation.

Also, there are a bunch of open positions for DOTS.

 

Those news was from may according your link and Unity say that there will be no updates until end of yer because they work.
You know as a developer that stabilization is huge step and time consuming, so having year to do actual work is very very very right decision. If you read this carefully you can understand this https://forum.unity.com/threads/notice-on-dots-compatibility-with-unity-2021-1.1091800/

We there not fans, we developers I if we professional developers than we should understand this more than any one another. ECS is not like Early Access is more like first teaser of tech, like first (may be second) video about Cyberpank2077 back in 2012. Early Access is late Alpha or Beta while ECS is in Experimental or Pre pre pre pre pre Alpha stage.

It is very official actually. You can go to unity git repo and look at it.

You just very negative there

Yes exactly. They say no info in 2021 so there will be no info. If you want info - look for it yourself. It is Experimental tech.

Abandoned and silence is not the same at all. And you can see that there a lot of work in progress and many posts from unity staff about work and size of DOTS team around the forum. And Unity state clearly why and when. https://forum.unity.com/threads/notice-on-dots-compatibility-with-unity-2021-1.1091800/

This is just about professionalism. To do job and create what complaining community want. This is good step.
Just read forum. ECS community complain just about everything, mostly why your pre pre alpha experimental tech not super stable release, why new version of it not back compatible....
Silence is pretty good way to stop waste time with such community and do actual job.
Again dont use Experimental Pre pre Alpha tech. No one can say when it will be released. Unity was too naive when they set release date for DOTS 1.0. We understand this they too. So no more any release plans just WIP.

If you dont want to use Experimental tech just dont use it and wait for release.
After release future will be very rock solid. This for Every one not for you personally.
If you want to use such tech than you need read all stuff from DOTS sub forum to be aware of.

Sorry for so many text. Just want to write maximum answers may be this will help somehow stop topics like this.

I am not blame anyone. Just may be this new topic help any one and help stop creation the same topics again and again and fans (professional developers) already understand what unity wrote and that we need wait for 2022 and may be even wait for GDC 2022 where Unity say something for sure.

once again

ECS is EXPERIMENTAL

 

I'd be surprised if that link and that comment exists at all and nobody of us found it. Last thing we heard is that DOTS has the largest amount of people working since it started.

 

Doesn't matter if it can or cannot be found IMHO. It would be nice to address the original post if it can be though, and prevent further drama.

 

I don't really intend to participate in this argument. People are going to believe what they believe. But I do want to clear up some misinformation.

That 2.0 is V2 which is what we have and can enable today with the scripting define ENABLE_HYBRID_RENDERER_V2.

More like from May: DOTS Releases - Latest Release: DOTS 0.17

 

Not quite.
It is last version of V2 that is unavailable to us
ENABLE_HYBRID_RENDERER_V2 you will not have what is presented on SIGGRAPH 2021 because you dont have last version of package.
From their sample Current V2 can handle CPU side of rendering in 16ms while their last version can handle it in 5ms while current URP can handle it in 360ms

May be I'm wrong, dont reread papers to write those lines

 

I looked over the slides this morning, and I'm quite certain what I said is true. Here's the actual slide that details what is going on:


See the comment "(new code is even faster)"? That's the unreleased version. All the numbers on that slide are based on what is available today. Also, the techniques they describe are exactly the behavior of V2 which is quite different from V1. V2 actually has parallel GPU upload jobs relying on change version numbers, while V1 doesn't. That was the whole point.

 

Can you share the link or the name of the talk?

 

http://enginearchitecture.realtimer...c2021_unity_rendering_engine_architecture.pdf

 

I haven't seen it posted on the forums yet but this PR from the graphics repo was discovered by (I assume based on the username) @thelebaron and shared on the unity discord.

https://github.com/Unity-Technologies/Graphics/pull/5842

There is mention of DOTS 1.0, a new hybrid renderer, and entity picking support. Just more evidence that work is very much under way on DOTS/ECS.

I'm now worried that after sharing this, unity is going to crack down on what the graphics folks include in their PR and commit messages to avoid any mention of DOTS or ECS... I'm just glad to see some sign of progress being made and appreciate the limited visibility we do have into what is happening.

 

Looks great so far, I think we can all assume Unity is still busy at work.

 

It is very professional if you are not in liberty to disclose the source of the information.
Ranting arguing and calling people names about it isn't though.

Again, thank you for the link. That was all you needed to share in order to stay professional.

 

It is the same link the Lurking-Ninja shared.
Apparently, it has been received and "translated" differently by some people.

 

Does anyone know what the entity picking support is about?

 

I would assume: being able to click on objects in the scene view & have this select the entity and bring up its inspector

 

Huhh?

 

There are so many of these posts around.
I think we can all agree that Unity is doing a terrible job at developer relations for DOTS.
A massive company like Unity could easily spare a single resource to give a weekly summary of the goings on around DOTS. It doesn't have to be committal, it can just be a "hey how's it goin' DOTS devs!?".

The facts is that development is ongoing.
Evidence to support this:
- open job positions
- devs responding to forum posts
- DOTS is a gargantuan project
So don't be afraid about abandonment.

I'm starting to get upset at how bad Unity is handling the PR of this.
Big names in the Unity developer community are getting concerned and enthusiasm is waning for many.
From a purely business stand point it makes no sense to ignore your early adopters and advocates.

TL;DR Unity give us some regular updates please!
We don't need to know the development schedule, we just need to know you're there.

 

Oh...again..."DOTS dead\abandoned"... I'll just sit around and look if you don't mind...

 

It's nice to hear that. Thanks for the heads-up!

 

Thank you so much for the update. Can't wait for more info about DOTS future and current progress.

 

Maybe you can cooperate with the community, there are still many smart people in this world, just like unity assetstore

 

It might be be dead, but what are the chances that the current iteration of DOTs is effectively a dead end, and this new DOTs stuff is incompatible with the old DOTs?

 

Until something leaves alpha and beta, you can expect change. I hope this is not surprising.

 

Extremely unlikely for several large parts of DOTS that are already being tightly integrated into various parts of Unity (Burst compiler, Job System). Almost guaranteed for some parts that were already scheduled for a rewrite for a long time (Cecil-based codegen to be replaced with Roslyn source generators - sounds like a ton of work tbh, no wonder it's taking a while).

As for ECS, it took them quite a while to nail down the API so I would be surprised if there were too many breaking changes, but it will only be harder to justify them after 1.0 hits the shelves, so it's possible they'll take the opportunity.

 

Sometimes I have the feeling that some think DOTS == ECS, and forget that DOTS is more than just ECS.

 

It's basically the blob that is Unity 2.0 vs today's monobehaviour-gameobject-mix

 

At this point, DOTS is a new game engine. Unity should consider letting it fork off, release into the wild for a few years, then consider selling their least favorite engine to an entity like Facebook.

 

Ok, let's push you're idea further. Unity was originally called Unity3D. Do you think they should have forked it to make Unity2D ? Do you think they should have forked it when they made Shuriken ? or when they made the Animator controllers ?
Why can't I chose a fork of Unity where there's no physic engine, or no terrain engine (after all, I never used the Unity terrain) ?
I think forking is not the solution.
Unity is a game engine, it is supposed to offer a broad set of features, possibly with multiple alternatives to achieve a given goal (like we have Animator and legacy Animation and Timeline). Some of its parts will be deprecated over time, sometimes, when it has become a burden or is perfectly replaced, but the goal is not to have only one way of performing a given task.
Having multiple way of doing essentially the same thing is a strengh because it means more workflows can be supported in a tight and precise way.

DOTS and GO both serves the same purpose, and solve it in a different way. Maybe ONE DAY, in the FAR FUTURE, everybody realises that DOTS is so good and so easy to use that GO is not of any importance anymore and we can drop it. But this is not the day.

I personnally would be very disappointed if they'd fork GO and DOTS.
One reason is that I don't expect my studio to go full dots anytime soon (even if DOTS 1.0 stable and all was released yesterday, there's still a big adaptation cost), but if DOTS and GO interface properly together, we could imagine to have a GO project using DOTS here and there for some specific part of the simulation (mostly, say, complexe visual, I don't know...).
This would become impossible if they fork the two into different engines.

What might please you could be to eventually put GO into it's own package thus making it optionnal. With, say, another package being called "Dots/GO interface" to bridge the two with relevant API if necessary.

That way you just could install the packages you want, and you'd have you're pure dots engine you dream about. And I would have my hybrid engine that I hope to have.

 

Me waiting for DOTS to get verified:

 

Joachim resinured me with his short memo here upper. Anyway, DOTS cannot be abandoned as the performance of Data Oriented programming is so high, on which other engines also work on, that Unity would be eventually die / performance is everything.
This lack of communication is the killer, and made me also worried, so thanks to Joachim for recalling the obvious future of DOTS.
Just look at that if you have DOubTS: https://forum.unity.com/threads/share-our-dots-showcases.1010215/#post-7503938

 

So I guess I'm gonna break my "don't post on the forums" rule here, but I gotta ask

Why should anyone take this seriously, given Unity's long-standing issues in this particular field, especially when long periods of silence and "we're sorry for our communication issues" are both things that preceded all sorts of other dropped features or features that saw massive internal reworks that lead to substandard releases, like with the canvas UI tools?

If the purpose of the package system and the preview/prerelease/whatever it's being called now was to provide people with ways to engage with Unity as far as package creation is concerned, it feels like that experiment has had mixed results at best. If the purpose was simply a Unity test bed that allowed for lots of data collection in terms of use habits and one-way feedback, however, maybe it's more useful on that front.

The problem remains, however, that "we'll have an update soon, promise!" is one of those things that gives both long-term users of the engine and relatively new ones pause for good reason. It's reached the point where the only context I ever use Unity for anymore is prototyping because of the state of the engine and its ongoing development, and that's hardly a unique standpoint anymore.

There needs to be better communication from Unity across the board, but there especially needs to be better communication when a package has been frozen to a specific Unity release where there's been no substantial updates for nearly an entire year with communication basically being yet another thread where Unity drops "hey we're not gonna be doing any major releases for a while" back in April that got a handful of non-answers from people asking what's up with that a week later, and then silence.

I don't think it's at all unreasonable for people to want a real answer as to what's going on with DOTS and for people to start assuming the worst, especially with Unity's particular history in terms of both deprecation and communication.

 

Crypto is transitioning to WASM with ETH2 in Q2 2022, should I consider thinking about DOTS again?

 

Disclaimer - I'm just an amateur and don't work in the game industry.

I'm under the impression that DOTS is at the very least considered to be a future selling point of the engine. It's an important technological direction in which it currently has a clear advantage over UE and that's a pretty big deal from a long term business perspective.

So I find it pretty hard to imagine that it would get abandoned at any time in the near (~3 years) future - it's very much in their interest. There could be issues and maybe it will never be fully finished, but I doubt it would happen due to lack of effort.

 

I don't care to be honest. I am enjoying current monobehaviour Unity. Unity took care of the existing users with gameobject networking and so on. For as long as Unity doesn't abandon existing Physx/mono Unity and even improves it, DOTS can take as long as it wants. It is alpha / bigger team stuff after all.