So tired because regressions after each update !!!

This made me laugh, 2021.3 LTS has the most unstable releases I've experienced in 5 years of using Unity.

An absolute highlight is a random removal of a core feature in 2021.3.9 https://forum.unity.com/threads/rec...ion-is-missing-or-better-say-removed.1335368/

 

The UPR is great... when it works ^^
Mainly to save on drawcalls and performance in VR

 

HDRP sample literally dosn't compile on latest 2022

 

Due to package errors, the compilation is not performed.
But to use some words that I have read a lot in recent years:
"we were unable to reproduce this issue, we need you to submit the project with the bug."

 

Still on built in for this very reason

 

Yep. I tried starting an HDRP project and had to use the other HDRP template to even get it to compile in the first place. The fact that made it to release is an absolute embarrassment.

 

That's not what I said.

 

Yep, sorry about that. I don't have good English and it's been a long time since I've written in English. I have a correct technical level, but not literary. Sometimes my lack of vocabulary fails me, I apologize. I only understood it after the fact, but it was too late to correct. I didn't mean to be unpleasant.

 

Unity tech team should stop chasing on features and switch to bugfixes for at least a year or two. Your goddamn product is unussable.

 

I and others have been saying this for over five years now but it's pretty clear it's just never going to happen. There needs to be a major focus on bugfixes, getting things that are already implemented into a proper production ready state, and dedicating work to things that have been in production for years. There needs to be a feature freeze, but that doesn't look good to investors or shareholders, so...

 

Well, whatever they are doing now also doesn't look good to investors and shareholders, so maybe trying something different would benefit them.

(and yes, the whole market is going down, but they're underperforming it).

So, who knows, announcing that they're getting their S*** together in efforts to become a really easy and fun to use engine that anyone can use might actually do wonders for the stock price.

But the problem is that what Unity needs is not "just" bug fixes.

Its features don't work well together and it's painfully easy to see that each feature was developed in isolation.

The new Unity features, besides being incomplete and janky, also introduce different paradigms, looks / feel and conventions for each one. There is no common design and no common way of working, no knowledge that the user can transfer from one feature to another. There is no "Unity way of thinking" to "get".

Also, each feature treats the user differently. Some are like actually just an API and the "feature" is just an example, and the user is supposed to figure out and code the solution they want, while others are completely restricting.

This even happens within the same features sometimes, when one aspect is "oh, you're supposed to code that yourself" while another is "this is the only way to do this and it's really restrictive, and there is no other way".

So even if they freeze the feature set and focus on bugfixing and maturing, there is no changing that Unity feels like a frankenstein's monster and going from feature to feature and tool to tool feels super jarring. <- And I don't see how they can fix that without pgp wiping their code base and starting from scratch and actually having someone make design decisions this time.

There is no saving Unity.

Even if they honestly start to improve things, the negative momentum is too strong and will continue for some time, user sentiment will continue getting worse and then at one point people will be ...done.

 

Well, from the current position those two things are directly at odds with each other. Maturing what they've got means they need to stop changing it. But getting things to feel consistent requires that they pick a standard and then change (read: "deprecate and replace") things to match it. They can't do both at the same time, so they'd have to do one and then the other, which will take years.

That said, I'm personally not super worried about things not being consistent. They aren't in most dev environments I've worked in. I'd love it if they were, but the fact that they usually aren't just doesn't seem to significantly impact my productivity. Or, perhaps more accurately, they don't impact my productivity as much as other things which are probably far more important, such as certain bugs which just... hang... around.

Also, in some places the attempt at consistency has been costly and painful. UGUI attempting to do things the Unity Way with GOs and Components made great sense at the time, and in some cases it works really well, but in others... well, I can't wait to try the new Flex based system, and am specifically looking forward to the fact that it uses standards developed elsewhere, which happen to not be consistent with The Unity Way.

 

I'd say this didn't make great sense at the time. Loads of people, myself included, took great issue with how it was being set up because it didn't match any existing UI workflows at the time and its reliance on GOs, components, and the hierarchy also lead to massive performance issues for years, as well as general scalability issues the moment the UI got even a little complex.

 

That was my point as well, which is why I concluded things cannot become good.

But in any case, they are not going to do either of those, so those options being at odds with each other doesn't matter.

Was it? I consider UGUI the start of "let's F***ing do whatever". I don't think there was a decision or attempt to implement some "Unity way", the design behind it was "let's reimplement NGUI, but worse".

 

Right, but NGUI was very much done The Unity Way, i.e. "make everything out of nested GameObjects and components".

 

UGUI is fubar granted, but at least it has some view / logic seperation. A complex UI is unmaintainable in NGUI but can work in UGUI. Im not protecting UGUI in any way or form but saying its worse than the legacy UI is stretching it too far

Edit: aha, i meant the old legacy OnGUI implementation not NGUI. it's called IMGUI apperently.

 

I think we disagree in "The Unity Way = nested GameObjects". What other feature required making a mess of nested gameobjects at the time?

 

Even in those days, you had point releases. 5.1, 5.2, etc. We had some pretty large features ship with those point releases in the 3.x and 4.x cycle (Like UGUI).

I think the bigger problem is that this release cadence (for paying customers) is at odds with the licensing model.
The only way to guard against these regressions is to not adopt major LTS versions right away (you can completely forget TECH stream exists for long-running projects). Meanwhile, you're paying a monthly fee and getting none of the benefits.

 

.
Before ngui and ugui there were GUIText and GUITexture elements. And a minority of people asked for a styling markup like CSS to lay them out and script them, which many people were upset by the idea of and shot down. This was before recttransform and it was a pain in the ass to work with.

So we went through ngui-> ugui and 3rd party textmeshpro -> unity textmeshpro

And then come full circle, 10 years after initially proposed, with the new ui toolkit with visual elements and USS styling more familiar to people from web world.

 

Not just the web world, but existing tools like Figma, NoesisGUI, and Scaleform. Having a markup language backending things and separating concerns has been a standard in gamedev UI tools for ages and Unity specifically bucked that trend by introducing an absolute nightmare of a system.

 

Well, typically you'd try to avoid the "mess" part, but physics would be an obvious example.

 

All this is pretty, but the final version has not even been deleted at the risk of sounding like amateurs.
At best, the final versions are Preview versions (very far from a true final version)

The best reason I see for having such buggy builds is that they take the community for their testers. Cheaper than having a service that takes care of the final quality of the product.

I had abandoned Unity for a while to switch to the competitor, but after so long exclusively on this platform, difficult to have the same level on the other platform.

But when I count the number of weeks / months lost due to their damn regressions,
I think that this time I will devote to training with the competitor.

And, finally, I'll say goodbye to Unity.

 

As a user of unreal I can say that there are regressions there too, and pretty much at the same frequency. Theres a lot of good reasons to use unreal over unity, but regressions (unfortunately) isnt one of them

 

I don't have enough perspective on Unreal at the moment to know that. But the frequency of updates seems to be lower, and especially when they release a new version, it is not completely unusable unlike Unity...

EDIT :
Also, the Unity devs can never reproduce even the simplest bug, so I don't see a bright future for it.
There are surely regressions in the competition, and that's normal, but on Unity even if one day they are fixed, the next day they come back. And they accumulate them so much that like this version, it's difficult to correct everything later...

 

It's just not regressions for me, it's the feeling that even the basic stuff across the board doesn't get tested before release. Some examples i've been hit by in the last couple years:

- Remember that time when editor gizmos (even the built-in ones for translation/rotation/scale) were completely broken on all platforms?
- Time.fixedDeltaTime broken in mobile platforms (in a major release that supposedly "fixed" delta time)
- Graphics.DrawMesh completely broken in 2021...
- ...so tried Graphics.RenderMesh instead, however it crashes the editor on all platforms when exiting play mode, reproducible 100% of the time. Also crashing when passing a mesh that has been destroyed.
- Graphics.Blit breaks the editor UI
- Built-in cloth has been deeply broken for years, both runtime and editor tools. Not sure about its current state, stopped caring.
- Raycast command only ever returns 1 hit, regardless of max hits parameter.
- Default List rendering broken in inspectors.

These are just a few, lately it's fairly rare for a week to pass by without us hitting some issue. As a result, I'm afraid of using any new APIs/systems in production regardless of how "ready for production" Unity makes them out to be, instead try to stick to the absolute basics (which also break from time to time, sigh).

 

I totally agree with you.
The FixedUpdate() issue goes back so far that I even stopped thinking about it on mobile. Contenting myself with the simple Update().

 

2022.1.22f1 is Out !

Seeing the Change Log of version 2022.1.22f1 and the short time between the 22 and the 21, I admit I hoped that a lot of our problems were solved.
Maybe I was too optimistic... In the end, the SRP packages problem was solved, but not the Additional lights issue on Android, and I wasted no time looking for other bugs...

Looking at the known issues, it looks like they're not doing much on this release, and since I started this thread on Oct 15, 2022 and sent in a bug report, a month later, I thought the problem resolved...

I hope for some, that they will find their happiness there and that this thread was used for something.

The next version will be decisive for me!

With just this screenshot, I think the worst developer could reproduce this bug... Maybe that's too much to ask, and it's not even in the known issues...
Unity 2022.1.22f1 + URP + Forward + Android



Nice legacy for the next major release, if he ever finds the problem...
Unity 2022.2.0 b14 + URP + Forward + Android

I don't know if it fix your Spatial audio issue but :

• Web: Added FMOD audibility stubs for WebGL. (UUM-15263)

 

Solved in september : https://forum.unity.com/threads/cas...al-audio-3d-not-working.1266461/#post-8445179

But i stay away of these version at the moment, i stay in 2019.x.x

 

This is what employees of a company say when they don't know what to say and they know that there is a problem.

 

This is a new regression in 2020 LTS (LTS mind you).

GI dies. This is how it looks when GI dies.



After restart GI is correct

 

There are forums and methods for filing bugs or solving problems. This has turned into just a string of gripes and reposting bugs posted elsewhere. Closing because it isn't going anywhere. If you have a bug, file a bug report. Post in the relevant forum to see if are solutions, if you are doing it wrong or if there are workarounds. Please report in the proper place to ensure it gets to the right people at Unity to look in to it.