Dear beta users,
We would like to invite you to tell us about your gut feeling regarding the current state of 2019.2 (b7).
A gut feeling should NOT be based on data or bugs or repro projects, but on how the product feels in your hands.
In addition to the poll at the top, here are questions we’d like your answers to:
Are you using 2019.2 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 2019.2 today? (Please remember to submit bug reports for the issues you encounter and reference them when you post about them here.)
How did your experience with 2019.2 develop during the alpha/beta phase?
How would you feel if we released 2019.2 in its current state (b7)? What makes you feel this way? Please elaborate.
I would be furious! 2019.2 is absolutely unusable!
I would be disappointed. 2019.2 does not feel ready.
I would have mixed feelings. 2019.2 is good enough to be used, but not great.
I would feel positive about Unity’s quality standards. 2019.2 is in good shape.
Ship it already! 2019.2 is a prime example of excellent software quality.
Please note that Visual Studio issues you might have been experiencing due to the Visual Studio Editor package will most likely get resolved with an upgrade to the upcoming 2019.2.0b8 since we’re [reverting the move of the support into a package for now]( Visual studio 2017 no longer opens in a13 page-3#post-4659089).
I currently am not using 2019.2 in an active development project, but I was really hoping I would be able to. The problem is that I just spent a lot of time making my own toon shaders that depend on a custom node to properly track lighting in the LWRP. Aside from a bug with the ambient node in 2019.1 (Reported) still exists in 2019.2 (from what I can tell) I have it where I want it. When trying to merge to 2019.2 I noticed custom node were gone. I also noticed that many of the distortion vfx were no longer working correctly in shader graph after updating to 2019.2. Yes, I did update to the latest version via the package manager, still remained broken. I didn’t dive to much into it following that.
Although, what I am really waiting on is the updates particle system, the bugs to be worked out in the terrain system (brush programming issues related to preset brushes), and a few other things I can’t seem to recall off the top of my head right now.
Currently, based on the needs of my project, I would not recommend anyone converting over if you’re interested in a toon style design as 2019.2 lacks custom nodes where are needed to replicate some needed lighting features.
Nope, just opened a few times a copy of my spare-time project to run some tests .
Personally I’m exicted for the TypeCache API , as it hopefully makes the editor a little bit faster again.
Otherwise mobile performance improvements sound nice, but I’m doing mobile at work only and we’re mostly on LTS releases. Thus the earliest we’ll benefit from those improvements is probably when we switch from 2018.4 to 2019.4 LTS.
Editor performance has worsen a lot of the years. For me, Unity 4.6 is the last version that actually felt responsive. From then on, each new Unity release caused the editor to get even slower.
This went so far, that I stopped working on my spare-time project, because the editor can’t catch up with how fast I want to work with it or how fast I actually used to work with it. I’ve to wait for the editor all the time and this is frustrating me, so I stopped using Unity at home and only occasionally do some performance tests.
A while ago I compared Unity 2019.3 to 4.6, but the same applies to 2019.2, so here are some of those links.
I’m disappointed with Unity releases or more specifically player and editor performance for several years. I’m probably disappointed with this release as well, because it’s near unusable for me. But on the other hand, this release doesn’t bring much to the table I need atm, so I can happily skip it and hope the next version is addressing my issues. So I picked option 2 from the poll.
1. Are you using 2019.2 in a real production? If not, did something keep you from doing so?
Yes*
2. What feature(s) are you most excited about?
TypeCache is actually the only feature I can name off the top of my head so let’s go with that.
3. What is/are the most annoying/painful part(s) of 2019.2 today?
Nothing
4. How did your experience with 2019.2 develop during the alpha/beta phase?
Had some early issues with an earliery alpha test (mostly with packages though). Went back to 2019.1 and only recently (b6) started using 2019.2 again and have had no issues since then.
5. How would you feel if we released 2019.2 in its current state (b7)
I have had no issues with 2019.2b7 so would be happy for release.
1. Are you using 2019.2 in a real production? If not, did something keep you from doing so?
Yes, we have to because we are using LWRP, and LWRP in 2019.2 is way much more stable than in 2018.3/4
2. What feature(s) are you most excited about?
LWRP 6.7
3. What is/are the most annoying/painful part(s) of 2019.2 today?
We are experiencing player crashes on 64bit windows, if we enable graphic jobs and Development build, the player will crash every 5 mintues inside GfxDeviceD3D11Base::SetupDefferredRasterState(void), we have to turn off graphic job or development build setting
There’s another thing we’ve noticed in 2019.2, if we open unity on a computer without internet connection. The performance of Editor is terrible, it costs over 50ms a frame inside EditorConnection.PollAllCustomMessages, and makes the editor impossible to use. However when I disable the network adapter once and reenable it, the problem is gone. But I need to repeat this time to time
4. How did your experience with 2019.2 develop during the alpha/beta phase?
We just migrated from 2018.3 to 2019.2, so don’t know how the beta was
5. How would you feel if we released 2019.2 in its current state (b7)
Beside the crash and the editor performance problem, others are fine
It would be very helpful if you could submit bug reports for these issues, if possible with reproduction projects. It looks like the graphic job crash is a known issue but it’s impossible to tell for sure without having a look at this specific incidence and another reproduction project would also help us verify potential fixes more securely.
Before releasing any of the stable v2019 versions, you teased a new UI, but so far it seems like that hasn’t been implemented yet. Is that still coming at some point, or has it been abandoned?
Are you using 2019.2 in a real production?
Yes. We are in a Quest production that we started when LWRP came out of preview, because our performance tests suggested a 4x performance improvement over built-in. Note that the production is heavily delayed because of the pain points below.
What feature(s) are you most excited about?
LWRP 6.7
What is/are the most annoying/painful part(s) of 2019.2 today?
Most painful is the limited testing that seems to have went into LWRP in Android VR. I opened lots of bug reports, each with different types of visual corruption, most of which would have been caught by proper automated tests of combinations of options. (I know the testing farm is being set up to improve this. But you were asking about painful parts, and this is painful right now.)
Now, this wouldn’t have much to do with 2019.2 if not for the fact that LWRP releases and Unity versions are so tightly coupled now. It’s really annoying that bugs seem to only be fixed for 2019.3 and rarely, if ever, backported to 2019.2/1 anymore.
I was already forced from 2019.1 release to 2019.2 alpha on this same project for the same reason, and I’m afraid it will happen again with 2019.2 (release comes out, I can’t use it but have to go on to 2019.3 since crucial bugs are only fixed there). Only reason I’m not going to 2019.3a yet is that not enough of the bugs have been fixed to justify such an upgrade.
How did your experience with 2019.2 develop during the alpha/beta phase?
See above. Some of the bugs I reported were fixed in 2019.3a (in a newer LWRP which requires internal Unity changes not happening in 2019.2), none were backported.
How would you feel if we released 2019.2 in its current state (b7)?
I would be disappointed. 2019.2 does not feel ready. I would wish for proper testing of different package versions against each other before releasing a Unity version. Yes, there is “verified to work with 2019.x”. But it seems combinations between packages and platforms break a lot, even if promoted as fully working. 2019.2 + LWRP + ARFoundation doesn’t work. 2019.2 + LWRP + Android VR has tons of bugs preventing us from releasing our product. USDZ package breaks at runtime with IL2CPP backend in 2019.1/2 (yes, it’s still in preview). I could go on (and yes, there are bug reports for these).
Man i thought i am alone with that. Honestly i feel so bad because of that. It lags, toggle window full screen is so slow. Starting the engine now is so slow as well ! Selecting many materials at once makes the editor so slow that it takes a minute to get responsive - this issue is since a few versions back. When editor comes on focus it needs a couple of seconds to get responsive, and i tried to use the profiled and it showed something about rebuilding UI canvases. This is every time when editor gets on focus.
However i tried the 2019.3 version from a link using the new editor UI and i think it was much more responsive. I did not test it that much though. But yeah, editor performance is one of the main issues i have now which makes the development much slower. 2019.2b7 also crashes constantly, like on every 10-15 play mode entries.
Not looking into new feature, mainly hoping for optimizations.
The editor is so slow and the scenes are starting/stopping really slow. It can take like 10 seconds for a scene with a small terrain with some trees and the HUDs available in the game.
LWRP and Unity Engine updates are too tighly coupled (again, teased way too much in advance so lot of people start using it early, and even now no backports whereas it’s out of preview, feels like cheat)
Editor is slower on each version, this one included (hopes for 2019.3 but still, worrying)
2019.2 feels more like a 2019.1.B, and 2019.3 feels like it should be 2019.2
Does not feel ready but again. Doesn’t feel like its gonna implementing any of the fixes I need in my first lines until 2019.3 anyway.
Honestly, I’d be happy to wait until you’ve fixed all of the major bugs. To consider releasing with 25 known issues that aren’t to be sniffed at, it just feels like you’re trying to rush. I’d rather wait and get a stable Unity release than one which is available but not as stable.
It feels about average for a beta at this stage. But it’s one of those versions that just doesn’t have much interesting in it. Likely we would skip it entirely and wait for 2019.3.
Are you using 2019.2 in a real production? If not, did something keep you from doing so?
Yes, because we found we got much faster compile times with 2019.1, but couldn’t move to that because of 1148469. Once that was fixed in 2019.2 we went to it pretty much immediately.
What feature(s) are you most excited about?
We were hoping that we’d be able to use LWRP for shader graph, but it has other issues with terrain it seems so we haven’t been able to yet. Other than that, any and all editor optimisations, in particular compile time.
What is/are the most annoying/painful part(s) of 2019.2 today? (Please remember to submit bug reports for the issues you encounter and reference them when you post about them here.)
Not being able to attach a debugger by hitting F5 in VS is annoying (see 2019.2.0b4 and VS 2019 ). But at least there’s a workaround.
There appear to be some issues with a missing component script in Rewired that I suspect is related to the UnityEngine.UI.dll problem. But not totally sure,.
How did your experience with 2019.2 develop during the alpha/beta phase?
We moved to 2019.2.0b7 without using any previous version of 2019.2.
How would you feel if we released 2019.2 in its current state (b7)? What makes you feel this way?
Wouldn’t be too bad, but at least without a fix for the VS issue it would feel like it was premature as that adds like 5 seconds to my workflow around 100 times a day which is a bit much.
Are you using 2019.2 in a real production? If not, did something keep you from doing so?
No, double clicking C# script and not open VSCode in the correct file and line is the deal breaker. The VSCode package plugin still doesn’t work.
What feature(s) are you most excited about?
Timeline 1.1.0 package potentially contains a fix that is a blocker for my game’s release. (Deterministic control track evaluation, and even frame number bug fix) but it seems to not arriving at 2019.1 for very long time.
I want to see the sprite editor package code and see if I could make something more.
Sliced UI pixel per unit will greatly increase my UI variations I could assemble from very few pieces. I already using duplicate texture only for varying pixel per unit. (for thinner border, etc.)
Screen.cutouts, UI design in games will have a lot more room to play with.
ClipEditor, TrackEditor, and MarkerEditor
I don’t do 3D games so other features are kinda out of reach. But new AI denoiser may be used for my hobby Unity-based music videos.
What is/are the most annoying/painful part(s) of 2019.2 today? (Please remember to submit bug reports for the issues you encounter and reference them when you post about them here.)
Double clicking C# script is not talking properly with my VSCode.
How did your experience with 2019.2 develop during the alpha/beta phase?
I can’t stay in 2019.2 because of VSCode problem. When I found a bug in 2019.1, I go to 2019.2 to reproduce it and sometimes found more. The I try something else in 2019.2 before go back to 2019.1.
How would you feel if we released 2019.2 in its current state (b7)? What makes you feel this way? Please elaborate.
I would have mixed feelings. 2019.2 is good enough to be used, but not great. Well the thing for me is only VSCode…
Well mainly it was that we’ve got a custom terrain shader and it would’ve been some work to get that running with LWRP. But even with the default terrain material and LWRP, with our setup there were issues with grass/details being culled strangely.