Feedback wanted - Your feeling about 2018.3 in its current state

Unity Engine
,
1

Dear beta users,
We’re closing in on our 2018.3 release target of early December and would like to invite you to tell us about your gut feeling regarding the current state of the release.

A gut feeling should NOT be based on data or bugs or repro projects, but on how the product feels in your hands.

Here are the questions we’d like your feedback on:

  • Are you using 2018.3 in a real production? If not, did something keep you from doing so?

  • What feature(s) are you most excited about?

  • What is/are the most annoying/painful part(s) of 2018.3 today?

  • How would you feel if we released 2018.3 in its current state (b11)? What makes you feel this way? Please elaborate.

  • I would be furious! 2018.3 is absolutely unusable!

  • I would be disappointed. 2018.3 does not feel ready.

  • I would have mixed feelings. 2018.3 is good enough to be used, but not great.

  • I would feel positive about Unity’s quality standards. 2018.3 is in good shape.

  • Ship it already! 2018.3 is a prime example of excellent software quality.

Thanks all, your feedback is much appreciated!

3 Likes
2

I voted 3.

No, because:

-There is a known issue about metal performance degradation, and I’m developing for mobile right now.

-Something keeps going wrong with skinned characters with animation. It seems they all get jumbled up in the scene and I need to reimport the mesh manually. This keeps happening even in the latest version.

-A lot of iffy stuff with reimporting meshes in general. Some objects complain about needing the normals fixed (something about blendshapes, even though I’m not using blendshapes for those), and when I hit fix (or change any setting for that matter) and apply, the lightmap gets destroyed, which is a problem for me, because if all lightmaps are gone, I need a week of baking to rebake everything. Or worse, the order in which submeshes get imported gets changed, so suddenly a submesh has another submesh’s transform/script etc. (<- This sometimes gets fixed by reimporting → reverting → reimporting again, but it means I have to check EVERY scene and every mesh to ensure that everything is still alright).

-Improved prefabs is something I have no use for, and relearning to do the stuff I need to do is something I don’t want to bother with right now.

Light Probe deringing. Aaaand, that’s it. Or maybe Legacy particle system being removed, just because it’s funny that you list it as a feature :stuck_out_tongue:

Upgrading the project is not smooth. Stuff keeps breaking. It has been the most un-smooth in a while actually.

Even though the last few releases were pretty decent, 2018.3 seems on par with what I’ve come to expect form Unity from all these years using it. Every release being a step forward, a summersault back and a bunch of sideways hops.

And at this point I don’t want to spent weeks fixing stuff in my project that were not broken just to have light probes that don’t look horrible.

Maybe all these are specific to my project for one reason or another, or I don’t know, but you did ask us how we feel.

2 Likes
3

1. Are you using 2018.3 in a real production? If not, did something keep you from doing so?
We are currently using 2018.2 in production, but after extensive testing we plan to push a 2018.3 b11 build to production next week. I am amazed with how stable this beta build is. We stopped using beta builds earlier this year due to instability, but decided to try 2018.3 b10 and was pleasantly surprised.

2. What feature(s) are you most excited about?
Pause the Garbage Collector (expected and happy with it)
Better prefab workflow (didn’t expect this to be so good :))
Faster compiling (didn’t expect this)

3. What is/are the most annoying/painful part(s) of 2018.3 today?
Every time I build, I get 2 ambiguous errors. They don’t affect anything so I would just classify it as ‘annoying’.

UnityEngine.GUIUtility:ProcessEvent(Int32, IntPtr)```
Edit: This one went away after a full system reboot for some reason!

```[Physics.PhysX] RigidBody::setRigidBodyFlag: kinematic bodies with CCD enabled are not supported! CCD will be ignored.
UnityEngine.GUIUtility:ProcessEvent(Int32, IntPtr)```
Edit: This one is likely happening on some prefabs I am not using in my 3 year old project, maybe even in a 3rd party asset.

**4. How would you feel if we released 2018.3 in its current state (b11)? What makes you feel this way?**
I would feel positive about Unity's quality standards. 2018.3 is in good shape.
This applies to Windows & Mac Standalone builds.
2 Likes
4

Your last two points are pretty spot on for me as well, @AcidArrow

Not entirely. Too much stuff breaks silently right now. I’m on the cutting edge on our branch, so I actually use 2018.3, but I’m not recommending migration yet. Still waiting to see smoother project upgrades and better support for the improved prefabs system.

I think work being done in the “backend” is the hottest thing right now. Compiler improvements, burst, non-alloc stuff, native arrays and friends, and, especially, the talk about getting the assembly reload times down in editor for better iteration times (still not much on this last point, sadly).

No matter how much you pack your engine with features, if the iteration times scale linearly (at best) with project size, too much time will be wasted there.

This has been, possibly, the largest pain point for us since Unity 3 and one of the major pain points for large-ish projects.

You can’t give total freedom to the creatives if each asset they add to the project adds a bit of time when entering playmode. Or when each type you define makes assembly reloads takes a bit more time.

5

I am using Unity 2018.3, with conjunction of ECS and latest previews.

I didn’t have chance to test many of new features, but since I am aware, many of them are in preview, I wouldn’t be complaining about it :wink:

Using on my main project and prototypes. No specific issues so far.
Had some small hiccup, but can not reproduce anymore.

For me ECS at given time. But looking later, to test some new graphical features as well.

Stuff in preview :smile: Well not really annoying, but continuous tension, keeping on tip of toes that something may change any time.

I completely don’t like UI Canvas system. Perhaps is because, I am a bit demanding on that part. But still feels somehow awkward to work with.

Generally I am happy the route Unity has took so far, in past year. So I would say generally yes.

Currently for what I am doing, I don’t see objections. But this may be very subjective, since I haven’t checked many features, which normally should.

1 Like
6

would be nice if gpu lightmapping would work on the mac, because I don’t think its a good thing to release a full version with features that only work on pc
it goes against the unity history of being cross plaform for development not only for builds

1 Like
7

Are you using 2018.3 in a real production? If not, did something keep you from doing so?
Yes because we dot have urgent releases and improved prefabs makes a huge difference when working on UI and other complex mechanisms.

What feature(s) are you most excited about?
Improved prefab but to be honest I didn’t took a very close look on other features since they were not needed or could wait.

What is/are the most annoying/painful part(s) of 2018.3 today?
Collaborate. Well, it was catastrophic before but since this beta, updating anything takes 10 min of “compiling” or something.

There are other annoying bugs like the F key not working and the package manager being broken. From what I understand these issues are specific of my project but I can’t do anything about it because they are all unity features that I don’t have any access to.

How would you feel if we released 2018.3 in its current state (b11)? What makes you feel this way?
I had to give the worst answer to that. I see from the other beta tester that 2018.3 seems stable enough but the issues I encounter make it unusable on my project.

8

Hi, tried to open our 2017.2.15f project in latest b11 and except 2 sigabrt when importing the whole project. ( I just reopened two times and it went through ), deleting DataPrivacy and importing new Playmaker for 2018.3 all seems to be working fine. :slight_smile: Since we are waiting for nested prefabs I feel like I can work with b11 onward but still will wait one more week for f1 release. We will start from there and make it our production version. M.

9

Are you using 2018.3 in a real production?

  • YES. (develop on and targeting macOS with Intel GPU)

What feature(s) are you most excited about?

  • LWRP and ShaderGraph are finally in a working state, releases for 2018.1 and 2018.2 have too many bugs.
  • Nested prefabs, prefabs mode (editing) are very nice tools to have.
  • Bugfixes that are relavent to our workflow.

What is/are the most annoying/painful part(s) of 2018.3 today?

  • Seeing LWRP is now stable for 2019.1, but not sure if it would be supported in 2018 LTS.
  • Few asset authors are ready for LWRP at the moment.

How would you feel if we released 2018.3 in its current state (b11)? What makes you feel this way? Please elaborate.

  • I would feel ok about it (VOTED 4).
  • For a beta, it appears surprisingly stable, and I have seen few bugs that prevented me to do my work.
10

This is because it’s not allowed to have CCD (continuous collision detection) on kinematic bodies. It was never supported at all, and in some cases led to issues. Now, in 2018.3 we have the physics engine upgrade that offers Speculative CCD that actually works with kinematics. I’ve received plenty of complaints about the mentioned error so we’re rolling out a fix that will switch the ccd mode to Speculative every time you set CCD on a kinematic, and will report a warning that explains what happened. One should still be conscious about the CCD modes though - some unintended suboptimal performance numbers might result from this.

Anthony.

3 Likes
11

Did you submit bug reports for these issues?

It’s important that you report these issues together with (a) reproduction project(s) if you want them to get fixed. If you don’t do it, it might be a long time until someone else reports them or until we find them.

The GPU Lightmapper is still in active development and was included as a preview feature in 2018.3. Mac support has been added in 2019.1.0a11 (the feature is still in preview though).

No, LWRP is still in preview and probably won’t receive bug fixes in 2018.4.

Thanks for all your feedback so far! Please, keep it coming.

2 Likes
12

My thoughts are that because of the fundamental changes between 2018.3 and 2019.1, its imperetive that we get 2019.1 into peoples hands as fast as possible.

Id rather we ship 2018.3 and move focus onto 2019.1, given how stable 2018.3 is as a beta.

In fact, we are considering moving from 2018.2 to the beta for 2018.3 given how the usability is so much better (prefabs anyone, amiright?)

2 Likes
13

I understand.

But as I’m sure you also know, creating reliable minimal repro cases and narrowing down the issues takes time, especially since it doesn’t seem to happen 100% of the time. And I’m fairly sure that low effort bug reports like “stuff brakes sometimes” with no reliable repro are not very useful.

So I wasn’t going to make a fuss about this, but you asked how we feel, so I replied :slight_smile:

14

It’s no less stable than 99% of your older releases. I suggest a shorter beta cycle now you have a clear alpha. As Unity does its fixes in Alpha then ports to beta, I feel beta should be regulated to just “bugs Unity missed” and soft-release basically.

Anyone disagree with this?

You can probably define this a bit better as then people will be more likely to place feedback where it’s actually going to have impact: on alpha.

By the time people have a moan about a feature in beta, it’s too late. So beta being a soft release in general (with bugs being possible) makes alpha have an actual point.

Currently alpha is borderline pointless while beta is such a strongly promoted draw. Letting people understand that alpha can affect the outcome while beta is mostly just bugs, will probably give both releases a point. Currently it feels mildly imbalanced, perceptually at least from this side of the fence.

On another note: Unity should probably stop using the term “preview” outside of package manager, and only have the occasional “experimental” unity build. Using both terms for Unity builds has led to confusion, so much so that some people use experimental builds as a base to work from and then become upset when this experimental build is majorly behind alpha and beta releases.

So making experimental only for editor builds on occasion and preview for package manager only, will likely just clear up a lot of fog. You probably should have the experimental builds have a text overlay in the editor warning that it is experimental and there are no guarantees. Seems people really didn’t get the memo at times.

OK experimental and preview are different things and still apply to Unity editor but the rationale behind my suggestion is that for Unity’s userbase which has wildly different competencies and varying grasp of English language, you should simplify it down to:

Preview: package manager

Experimental: standalone unsupported editor builds (which can also be preview, just not in name!)

It’s clearer for people with varying development and language competency.

–

Sorry if post got a bit long!

6 Likes
15

Not to be pedantic: since 5.1.0, LWRP is no longer behind preview flag. (regardless, it’s also only available for 2019.1; 2018.3 currently use 4.x-preview branches, due to core engine difference AFAIK.)

https://github.com/Unity-Technologies/ScriptableRenderPipeline/blob/master/com.unity.render-pipelines.lightweight/CHANGELOG.md#510---2018-11-19

https://github.com/Unity-Technologies/ScriptableRenderPipeline/blob/4.2.0-preview/com.unity.render-pipelines.lightweight/CHANGELOG.md#420-preview---2018-11-16

I have made argument that LWRP should be supported in 2018 LTS even if with reduced feature, so I won’t repeat it here:

https://discussions.unity.com/t/664083 page-21#post-3873364

https://discussions.unity.com/t/716433 page-3#post-3870511

16

It’s not going to happen and I know this upsets you but you have a concrete and belligerent answer from non-unity staff here to make it clearer. Sorry.

What you can do is diff it and port it yourself. That might well be possible. It would cost Unity the same amount of money to do it for you, and you’re the only one asking for it.

If you have any issues porting the feature you want, THEN Unity can help! They just can’t choose to backport something that was clearly labelled preview for an older version.

But what specifically is your issue? Why not let people help you backport the bits you need (please post a new thread on this btw not here).

17

Yes, you’re right. If it is an edge case and hard to reproduce, its chances are not good. Judging from your description I wasn’t sure that this was the case.

The length of the beta phase depends on the number of unresolved high severity issues. One of the conditions for release is that all known bugs that are considered shipstoppers are fixed.

3 Likes
18

Just to be clear:

  • I am not asking Unity to backport any feature from 2019.1 to 2018 LTS. These 2 branches have already diverged (4.1.0/4.2.0 were released after 5.0.0 to fix some critical bugs we reported).

  • I am saying it’s bad look for Unity to create so much enthusiasm for LWRP during 2018 TECH cycle, only to end with “well, thx, but you have to use 2019 TECH release for LWRP to be supported.” (And it creates a weird dilemma for us where we can’t trust Unity’s 2019 roadmap, but we have to because LWRP is on the line.)

  • I don’t want to go on a quest to ask twitter/reddit to support us in “keeping a Lite-LWRP for 2018 LTS”. But I believe the need exists, if Unity team want me to do that to present a business case, I can.

Sorry to sound so negative, even though I generally enjoy working with 2018.3b and LWRP (when they work, they are fabulous.)

4 Likes
19

Hi AcidArrow,

This issue should be fixed now. Are you encountering still performance problems?

Marton

20

  • Are you using 2018.3 in a real production? If not, did something keep you from doing so?

  • No, I was having an issue loading external assemblies. Which in retrospect, I think was an issue on my end.

  • What feature(s) are you most excited about?

  • Prefab workflow. It’s been a long time coming.

  • Package manager stuff, specifically Git support. External libraries has always been a major downfall of Unity IMO. But after the recent runtime upgrade, and planned package manager support, I’m pretty excited that Unity is finally addressing it.

  • What is/are the most annoying/painful part(s) of 2018.3 today?

  • The premature deprecation of UNET is the biggest issue I have with it. Many people have already aired all the grievances I have in the deprecation thread, I don’t think retreading it here is necessary.

  • How would you feel if we released 2018.3 in its current state (b11)? What makes you feel this way? Please elaborate.

“I would be disappointed. 2018.3 does not feel ready”. However, I think it should be released anyway.

To elaborate, I’m worried that 2018.3 is in danger of being delayed (just a “gut feeling”). If that is the case, I think delaying 2018.3 would impact 2018.4 LTS, I think it’s more important to ensure a proper LTS release, even if 2018.3 isn’t perfect.