Official Notice on DOTS compatibility with Unity 2021.1

What is this comment? Using the DOTS stack is utterly different architecture than mono. You're talking about a full game rewrite. Also the reason I use this new stack for my current project is because of the monstrous performance gains unachievable using monobehaviours.

 

We are in hybrid so if Unity were to trash DOTS, of course it would be a non-negligeable amount of work to revert, but it is certainly doable in less than two months. That is why knowing if they are sticking to it or not is so important to developers, there is always a way out of an unfortunate situation given we have the time to do it.

Not all devs are using DOTS in its pure form to create games where you fight one million soldiers or to deform terrains. We still use legacy Animator, Physics, AudioSources, Canvas and so on. The Camera isn't even DOTS-friendly. For our game's prequel, we were using dependency injection (Zenject) and thought now that DOTS was heading toward something much more interesting for the future of our company, so we adopted it to learn and evolve with it. Separating data from logic has more benefits than just performance.

We decided to adopt and evolve with DOTS because of its long time benefits, not because we wanted instant monstrous performance gains.

 

Once you really start using DOTS and doing stuff like deterministic or network roll back, hybrid becomes untenable. In these cases you can do hybrid only for graphics.

 

Sounds like a bad idea for any non production ready package. You may end up waiting months for Unity to fix bugs. Use anything in preview only when a project could benefit a lot.

 

Pretty brutal that something so hyped might be the thing that distances OG Unity users away from newcomers, due to the controversy caused by DOTS/ECS integration and the initial "Hybrid" aspect of it.

However I still believe Unity as a team has big hopes for ECS, it's just a matter of not sweeping away their clientele by lime-lightning it so much anymore.

 

"Production-ready" doesn't mean anything when we are talking about Unity. Current LTS only recently become barely usable, with dozens of bugs fixed in the TECH stream, but not yet backported. If fact, ECS is 10 times more stable and "production-ready", than LTS Editro itself.
For context: https://forum.unity.com/threads/2020-3-8-l-t-s-is-completely-broken.1109702/

 

Our project started in 2019 and back then DOTS Core was supposed to ship in 2021. Also back then HDRP was not production ready. Also back then we were not in production; we were prototyping.

A roadmap serves that purpose: Projections. So we projected it would be safe to adopt DOTS and HDRP as we thought they would only get more stable. And it became all more stable. We logged bugs, they got fixed in a matter of weeks. Our game currently runs on PC, Mac, Xbox O/S/X, PS4/5 and Switch with no major issue except some scene loading crashes coming from the core of Unity, not in any package we have.

The whole point here is not to throw rocks at Unity even if they trash it; that is a risk we took and we assume it. The whole point here is that we want news that are not: "We just acquired a remote desktop app" or "Hey! We're heading to GDC with Lenovo to discuss nothing at all".

 

Sometimes it feels like Unity is being subverted from the inside by competitors, seems like it got substantially worse after it became public. Is there any company that didn't get substantially worse after becoming public?

 

Where did you get that date from? From what I remember it was supposed to ship in 2019 and in 2021 most systems (Audio, Animation, Physics etc) ported over.

Congrats for being lucky enough to use features that work properly. It wasn't the case for me with UNET and DOTS Netcode and I won't trust Unity in any future roadmaps.

Imagine what happens if Unity fixes it but only in 2021+.

 

Tbh I did not look it up I just wrote what I could remember but you're right that the core was supposed to ship in 2019, that only strenghten my position that it was a rather "safe bet" based on their roadmap.

I guess not doing multiplayer is a much simpler endeavour.

Well that would mean they do not provide "Long Term Support".

 

@MartinGram
Are there going to be actual releases between now and 2022? Or is dots just skipping the year entirely?

 

2022.1 beta/alpha near end of year?

 

I'm not a religious man, but let us pray for the beta release of entities.

 

Nor am I, but amen to that.

 

Oh dear Lord C++... or is it C# ? damn it, I get all mixed up with those prayers... or should we pray Saint Joachim ?

 

I've heard ECS was mentioned for 10 seconds during one siggraph talk, maybe it's not dead yet. That was the slide:

 

Do you know the name of the talk ?

 

https://cesium.com/blog/2021/07/20/open-metaverse-bof-2021/

 

So when you say dead.. besides not supporting the latest unity builds, are there any indications besides this that compromise unity's long term support of ECS? It's hard because it seems to me like a engineering issue and not a political one, but at the same time... it's hard to tell from the outside. IPO always leads to changes.

 

No answers for few months in this thread, no updates since 2020, no dots jobs listings, no dots on https://unity.com/roadmap/unity-platform

Anyway for me it doesn't matter if it's trashed or not. I'm more interested when it will be released/stable and the nearest possible Unity version is 2022.3. Hopefully I'm wrong.

 

The only in progress item on the Unity Animation Roadmap is "Advancing DOTS." In the Gameplay & UI Design section they have "Introduce DOTS scalability power to visual scripting." Here is a job listing for a DOTS role. Also, the latest patch updates to Entities and Hybrid Renderer packages came at the end of May 2021.

 

Nice, weird how they are not in packages change logs.
https://forum.unity.com/threads/dots-releases-latest-release-dots-0-17.1044523/#post-7185604
https://forum.unity.com/threads/hybrid-renderer-0-11-0-preview-43-change-list.1116457/

 

Shows up in the package manager.



My own wild speculation on a release of new versions of packages is that it will come either when 2021.2 leaves beta or when 2021.3 comes out (I think future DOTS releases will require compilation / scripting backend changes coming in 2021.2).

 

I imagine how things proceed once we have a new version of ECS published. Will the silence remain (since all that IPO/stock market rules still apply)? Are they planning developing this without the community? Just popping out a new version from time to time and reciving bug reports. Or will they go back to "normal" communication patterns? Both scenarios are pretty awkward. How will they attract new users (testers) when people see that they don't talk to their community? How can someone risk using ECS in a commercial product when those using it are left alone? Only people able to try it are those who just want "to play around". Is this a good testbed for high profile software?

Unity is a serious business and it should understand the business concerns of their users better than ANY of us. When you get no information you can't make business decisions.

And these 2 months cost you serious money. So you lose much money because UT refuses to say anything about what to expect. Practically you are beeing punished for using ECS and adopting a new and highly advertised "preview" technology. I doubt many people will make that mistake in the future. It's as if UT recommends us NOT to use ECS. Like a virtual middle finger. For me this move is incomprehensible for a company which relies on it's customers and community so much.

 

Well yeah, since may has there been updates to the package, I think due to the engine changes. I think hybrid renderer was broken for 2021.2, but entities package still works. It's just weird that there hasn't been updates to it since May. I find that stranger than lack of forum updates. Especially after 2 years of using entities. IPO or not, users of the package will find this very strange. I'd hope they didn't fire the entities team. Or move them into different teams.
From experience delving into the packages, other unity modules weren't really integrating entities packages into their backends? Animations still had most of the code relying on unity engine animation stuff. And audio package did their own thing. Visual scripting was scrapped for using bolt. I think the higher ups maybe thought it was too long term for them and their investors (they normally get lots of their income from shares and company assets). It went against what I expected at the time for those additional new packages, especially after the ECS presentations of 2018. Meanwhile Burst shows overall performance boosts as it doesn't need to be reworked into existing unity engine code and works at the lower layers.
I should also note: Entities is broken with il2cpp builds, so im waiting on an update to fix this bug.

 

Isn't it what UT does for about a year now? Recommending to not use ECS in production. This even extends to hide this package in the package manager just for ppl to not use it thinking it is "almost ready" because it's in preview.

 

When not for "serious" large scale projects, for what else should ECS be used then? Or who SHOULD use it at all? Is it really a feasible testbed when only small scale hobby projects or prototypes are developed using ECS?
And when @eizenhorn posted about his Unity DOTS case study in production there were no single official voice "warning" him or telling him NOT do use it for that. How risky it is or what can go wrong. So this seems to be a little two faced here by UT. Officially telling not to use it for production but when someones does it anyway they "applaud" him.
What is the expectation who should use Unity ECS and for what? It is absolutely clear that some people will adopt new and exciting technology for their projects. No matter what is said officially about its state and/or progression. UT now sits there and says "Haha, it's your own darn fault that you used our technology before it is ready". When we can't use it before it is ready why making it available in preview? The reason is, to gather feedback about it. But when "noone" can use it properly before it is ready how can they recive feedback? Such a preview development is a give and take. UT gives early access and receives feedback, improvements, bug reports. Community gives those (and their time) and receives an improving product. But suddenly excluding the community from any development and implementing a closed door policy is just "scaring" and alienating.
People ARE already developing frameworks and assets for ECS. For example Latios Framework or DOTSNET. They KNOW it can change and they have to adapt their code to new API. But they certainly don't expect a complete halt (at least in the official visible part).
When UT treats their "early adopters" this way, how can they expect anyone want's to try their "new, exciting, preliminary, hyped, preview" stuff in the future? They are literally saying "when you use our stuff before it is ready you are screwed". Does a company want to have such a reputation? I would say no. That's why I'm so surprised about this behavior.

Edit: There is no problem in hitting road blocks, encountering issues, don't have proper staff and such things. When this is communicated most Devs would understand (I suppose) as most have similar experiences with their own projects. The point is to still let the community participate and don't exclude them.

 

There's always risk vs reward factor.

Currently there's more of reward than risk to use in production.
Since ECS is as stable as the rest of the engine.

Sure lots of systems that are out-of-the-box aren't available yet. But that's just mean you can roll your own project specific solution. And reap performance benefits from it.

Recommendation to "not use" is there to prevent wrong and subjective opinions on non-ready product.
Which Entities are. As well as to prevent major headache for providing support during development of the package.

Whether actual development is on, you can get some insights from the posts like these:
https://forum.unity.com/threads/unity-the-slippery-slope.1157363/#post-7425569

Most-likely UT cannot share anything for the reasons unknown. I'd suggest wait and see.

 

Plural? Can you link 3 more "posts like these"? From the last half a year ofcourse.
In this post, Tautvydas-Zilys says "They just wanted to develop DOTS" implying that he does NOT belong to the DOTS team. DOTS team itself prefers NOT to react/respond. Before that dreadened IPO there has been many "talkative" posts from DOTS related staff. It's unlikely that they all lost their voice at the same time. So it appears to be an ordered silence. Without explanation.

That sounds as if I would have a choice.

 

Yeah, they are, and I don't see them complaining about it on the forum.
Don't get me wrong. I'm not saying, UT silence is a good thing or that nobody shouldn't feel frustrated. Just near 7 pages of complaining about the current state of DOTS and various information policies isn't a sign of being a good community. More like being that child in a shop that cannot stop crying about things it wants.
Sorry for such harsh comparison, but nonstop complaining serves nobody. It doesn't force UT to respond and probably makes them much less likely to collaborate with such a community in the future.

You never had. Besides choice to use it or not.

All that said, the one is certain: UT said its piece, and probably stays silent until 0.18 or whatever version is next is ready.

 

Is the child paying its parents, per month?

Complaining about other people complaining is my least favourite form of complaining.

 

In the case of the creator of DOTSNET, that's because he's been blocked from expressing any more of his thoughts in the thread where he was critiquing things, and his posts were deleted.

There may well be other cleansed discontent.

 

from the DOTSNET thread

Sounds like a "complain" to me. And a righteous one too.

So a "good community" is one which just "swallows" what "authorities" dish out? Expressing "dangers" (frustration, "incredibility"), articulating "issues" (no planning, lost trust) and detailing feelings is "childish behavior"? There are many people in the world stating when something is wrong (metoo, strike, whistleblowers, NGO's, human rights movements etc.). We can't know in advance which of those attempts will be fruitfull and which not. Do you suggest to keep silent because it COULD be useless to speak out?

I guess there are some people frustrated and alienated enough to not want to "collaborate" with UT anymore. Meaning they switch to a more "communicative" community/engine/product/company.
If I would make a big mistake in some regard, I WOULD like someone telling me.

In the end, UT weighs the pros and cons of either decision. But if they don't know the cons are (ie. what "burden" they put on their users back) they can't weigh it properly. So from my (limited) PoV it is imperative to inform them what is happening here.

Why would a company annoy and frustrate their users up to a point they don't want to use their product any more? What I hear from large Tech companies they desperately try to increase market share, user retention, public reputation and in the end money. What happens here is quite contradictory to such goals. Don't you think? So maybe they have secret goals which are not interfered by such "behaviour"? But if they have those goals they are doing something wrong.

 

FFS, if I remember correctly he compared this situation to parents' whose children were kidnapped. That's not discontent. I don't like the banhammer, but that guy in this thread deserved it. And he wasn't banned in any other threads AFAIK.

 

What's worse - poor choices of metaphors or bringing them up out of context.

 

Ah, I see. You think comparing information deficit to situation parents with kidnapped children are in is

or

"Out of context". Okay then. Bye troll.

 

I don't know what he said. It's gone. Without context, or proof, why should I believe your claims that that's what he wrote, how it was written and what he was trying to communicate?

 

Without mentioning anyone:

Maturity means when you see a roadmap, you know the map is probably already out of date. It's a guide, which is not a law that's immutable and infallible.

The answer is no news, and it will remain no news until otherwise. Shrieking at Unity will not change that until Unity has more information.

So this is locked.

When Unity is ready, they can post.

 

Hi everybody, I am happy to share some new information about DOTS development over here!

In a nutshell, we are preparing to update a set of packages to version 0.50 compatible first with 2020 LTS, and later 2021 LTS. In parallel, we are progressing toward the first fully supported 1.0 version of Entities and several compatible DOTS packages.

Laurent