Unity 5 releases have been a mess

We already have this forum, nevertheless the average quality of a bug report of free users is lower.

 

I had a previous project crashing on startup with 5.2 as well. Deleting the cache (Library folder) completely fixed it.

 

Usually does, it not since Unity 5 checking the Editor log will usually guide you to the offending asset, often a shader, that you can delete in the file manager manually.

 

Thanks for the insight. It's always much appreciated; just wish it'd happen more consistently across the teams.

 

Which is IMO B.S. - if it crashes, it should be delayed until crash is fixed (of course with information why it was delayed).

 

My main issue with this kind of approach, is that the official public release stays in that "broken" state for a month. Even though the patch releases are coming and are fixing the majority of issues, the official public release is probably what's get downloaded by the majority of the users.
Though that might not be the case, and the official statistics might prove that 90% of the users are staying up to date with the latest patch release.

 

5.2 forced me for the first time in many years to go back to the previous version I was using.

 

This would be more useful to me if it stated the expected release version.

It is clear the the fixes have to go through a QA pipeline and any prediction might need to be revised but "future version" is about as non-specific as it gets.

 

I can't even track my report as I can't find it in the issue tracker (and I've set my report description as public).

 

As far as I know, they become public only after they were successfully reproduced.

 

Yeah, I know, I've got an email that the bug was reproduced

 

superpig alluded to the fact that late "fixes" introduce problems. So, delaying a release does not reliably make it more stable, and delays beget delays.

We've already confirmed that at least one of the crashes in f3 was caused by a fix during RC period. (since been fixed I believe for p1 I think) Halo effects are hard to capture, which is why we lock down the release and fixes in bound. It also gives us more time to make sure we get the right fix. We've also had the experience of realizing the fix bound for patch was still not correct, and thus had time to fix it.

Though, you do bring up a valid question about feedback. It's tempered by the other side of it, that we blogged about (http://blogs.unity3d.com/2015/08/17/the-great-incident-bash-of-2015/) that opening the floodgates does not actually get us better feedback.

 

Following this through to its logical conclusion, we'd never get updates or releases because developers would always sit on it until they'd fixed everything, and since nothing is ever perfect nothing would ever get released.

I mean, that's why we have updates in the first place. Because something might be "ready" but it's rarely ever "finished".

 

I think right now Unity's in a bit of a pickle due to so many new systems being added / changed. Think about it, it's not just fmod and physx upgrade. It's a wide-scale multi thread effort, refactor, renderer, optimisations, IL2CPP, and everything else on top (with a cherry), because Unity had no choice but to splurt all this out in one go now. I guess hopefully bugs will only get less from this point onward.

*raises a single eyebrow*

 

If UT can reproduce the crash they might hold it back a bit. But if it seems to be an isolated incident they probably will chance a release.

 

So, from my view point, we had less bugs entering 5.2 final than with 5.1 or 5.0. However, we're being more transparent about what problems it's shipping with. Also, my definition of "critical bugs" to fix is pretty harsh during RC. My view point is that if the editor is functional, individual players can be slightly broken (but hopefully have workarounds) - as users will shift to patches or updates when they need to ship.

Regarding shielding users, there's a question of threshold. When 60% of users may find everything fine, would you still ship? 80%? 50%? What's your line? How many people does the bug impact? 100%? 10% of 10%? We have to balance the situation and sometimes we might make the wrong call, but I've also seen cases where the fix was worse than the problem being fixed. So, it's not such a simple thing to say "shield the general user".

Valid point that it impacts our support pathway more. We're still improving our internal processes, architecture and more to avoid this in the future. However, right now we're still in the transition from the "big-release a la 5.0" to smaller cycles. There are bound to be bumps along the way.

 

I've read the blog post and I think that it makes a lot of sense to release 5.2.1 right after the first patch. Good call!

 

oh man I tried for a few days with Unity 5.2 but I don't think i can take this much crashing and corrupted files.. OMG !! I'm losing my mind and many hours of productivity.

I appreciate Unity moving full steam ahead with features but this was released too soon.

We really need that patch asap..

 

We are usually a couple of versions behind (some times many versions behind). Unless there is a pressing need to upgrade, we don't bother. And when do, we only upgrade to a version that passes all our internal tests firsts. As a team we have to be all on the same version, so any upgrade is done with great care and deep testing in a separate branch. Commanders is still in 4.7ish.

 

As a freebie Unity user I always upgrade because those upgrades usually involve me being more productive with less. I do have a copy of the projects that are upgraded saved off elsewhere though.

 

yeah I agree with you. Your way of doing things is the wise way but I've been taking the risk of rushing forward.. I must learn to chill the f.. out!!!

 

If it ain't broke, don't fix it.

 

lol I dont think that applies here hippo

 

I've had to roll my project back to 5.1 as well... I was encountering an absolutely bizarre set of errors where (only in release builds) UI elements in uGUI would randomly disappear and I would get spammed with 'D3D9: attempt to lock null vertex buffer error' errors. Attempting to change my project to use OpenGL just crashes the editor too. I guess there really is a bug for everyone in this release.

 

@superpig Would be interested to hear more on the decoupling efforts to help lessen the side effects of changes, or more on how you guys are scaling automation and testing for earlier detection.

 

I love Unity! But if it does not crash less than 5 times during my work day i go celebrate with Champagne

 

It seems a lot of people had these experiences. Upgrading from 5 to 5.2 made my UI go into a train wreck-level mess. Upgrading to 5.1.3 made it work as before.

 

Well p1 of 5.2 is out today. Try it again as it's bound to have fixes.

 

I can fully appreciate that Unity has a lot going on right now and that it's difficult to tackle so many issues, but 5.2 is far from the most stable of the Unity 5 series. To say otherwise is to basically look blindly in the other direction. Check out the Animation forum and you'll easily see that for a large majority of us, Mecanim has been completely unusable as a result of moving from 5.1 to 5.2. So to say (and I do not believe this is done deliberately) "I haven't seen any issues" or "5.2 works just fine" feels a bit like a slap in the face to those of us that have been here since the very early days.

Also, there seems to be some issues in communication here as there is a lot of confusion as to what was supposed to be fixed in patch x and what known issues still exist in patch x that won't be fixed until patch y.

 

Patch is out today, and the UI is still flat out broken on windows standalone in D3D9.

 

Unless new version doesn't have critical bug reports within a week (for me "critical" meaning data corruption/crashes) or it has some features I desperately need (such as physics improvements), I usually stay on previous version. I can work around some oddities like for example those new PhysX had brought in, but if data corruption/crashes happens, it's a no-go for me.

Anyway, what I'm trying to say is "don't be an early adopter of software that is critical to your business and/or livelihood. Let people who just like to play around with it for fun try it first and wait for nastiest bugs to be gone."

 

That means pre-alpha and should not be release to public in any way.

To OP: Unity devs said they are focusing on PRO users when comes to bug fixing, and most probable big studios requests. However, we don't know the real numbers who's pro and who's not so we can't say anything statistically if some bugs are sent by PRO users what affect a majority on non PRO.

Anyway, I like that they posted a strategy for releasing patches weekly, so that's more that welcome. I really hope version release and bugs fixing will move from snail speed.

Unfortunately, simple fixes in this latest patch are still missing.. I mean.. seriously guys, you read the forums, you see what problems are, simple thing like re-enabling tree creator number of triangles, or increasing the GI slider to allow us to use more that 100GB cache isn't a end of the world to do. This is just an example.

 

I think the current Unity 5.2.x UI problems are supposed to be patched to work in 5.2.1f0 but if not then I think it means we have to redo our UI to use new versions of Unity :-(