Search Unity

  1. Megacity Metro Demo now available. Download now.
    Dismiss Notice
  2. Unity support for visionOS is now available. Learn more in our blog post.
    Dismiss Notice

Is Unity 2019 3 0a3 Compatibility with 2019.3 version

Discussion in 'General Discussion' started by amayeta5, Jul 10, 2019.

  1. amayeta5

    amayeta5

    Joined:
    Jul 10, 2019
    Posts:
    3
    Hi I'm a games design student and currently using Unity 2019 3 0a3 at college but I want to do some stuff at home for next year but not sure which version of Unity I can use at home for FREE which will allow me to run anything I make when I go back to college.

    Any help would be grate.
    Thanks Ama
     
  2. Joe-Censored

    Joe-Censored

    Joined:
    Mar 26, 2013
    Posts:
    11,847
    Why is your college using the 2019.3 alpha test? That version has about half a year of bug fixes to go before it will be considered stable enough for general release. The latest/greatest version of Unity is 2019.1.9 right now. Unity 2019.2 isn't even out of beta yet.

    But to answer your question, you just sign up a free account for yourself with Unity and you can download whichever version of Unity you want. Though I don't know if Unity keeps its alpha/beta builds, like the one you are using, available for download after the release version becomes available. Starting with Unity 2019.1, the editor generally requires Unity Hub, so you just install that and it lets you install as many different versions as needed.

    Unity Personal licensing is free, with almost no limitation compared to higher tier licensing, and only has a revenue limitation before you're required to upgrade to a paid license.
     
    RecursiveFrog and Kiwasi like this.
  3. amayeta5

    amayeta5

    Joined:
    Jul 10, 2019
    Posts:
    3
    Joe-Censored i actually asked them that same question but my teacher got really defensive and said that I just had to use what they provided and to stop asking questions.

    My tutor is not very accommodating and would on even tell me how to get the version they are using.
    So thanks a million

    My mum is helping me look at going to another college .

    Thanks again. I am coming off this and try your suggestion.
     
  4. Ryiah

    Ryiah

    Joined:
    Oct 11, 2012
    Posts:
    20,946
    Just grab the latest stable release. Upgrading an existing project to a newer release is far easier than downgrading it to an older release. Unity Personal covers releases as far back as Unity 5 but unless you have a very solid reason to I wouldn't go older than 2018.4 which is the current LTS.
     
    RecursiveFrog and Kiwasi like this.
  5. AndersMalmgren

    AndersMalmgren

    Joined:
    Aug 31, 2014
    Posts:
    5,358
  6. Ryiah

    Ryiah

    Joined:
    Oct 11, 2012
    Posts:
    20,946
    All of the releases have at least one bug we know about.
     
    RecursiveFrog likes this.
  7. AndersMalmgren

    AndersMalmgren

    Joined:
    Aug 31, 2014
    Posts:
    5,358
    I know, so it's not correct to call it stable, maybe, just maybe LTS could be call stable. Then again we were hit with a regression bug in as late as 2017.4.27 or similar. You are never safe :)
     
  8. Ryiah

    Ryiah

    Joined:
    Oct 11, 2012
    Posts:
    20,946
    By that definition we couldn't mark any program in the wild as stable.
     
  9. AndersMalmgren

    AndersMalmgren

    Joined:
    Aug 31, 2014
    Posts:
    5,358
    Offcourse we can, programs that dont have regression bugs in stable builds. Unity is the only software were I have ever seen that happen in.
     
  10. Joe-Censored

    Joe-Censored

    Joined:
    Mar 26, 2013
    Posts:
    11,847
  11. AndersMalmgren

    AndersMalmgren

    Joined:
    Aug 31, 2014
    Posts:
    5,358
    Yes, but regression bugs are something else.
     
  12. When do you have your birthday? I swear I'll ask the forum members to chip in a T-Shirt which says
    "I reject your reality and substitute my own" (Adam Savage)
     
    Joe-Censored and Ryiah like this.
  13. AndersMalmgren

    AndersMalmgren

    Joined:
    Aug 31, 2014
    Posts:
    5,358
    A regression is when a new feature or bugfix breakes another part of the software. If you have automated test in this case on the ToggleGroup component that regression could never have happened.
     
  14. amayeta5

    amayeta5

    Joined:
    Jul 10, 2019
    Posts:
    3
    Thanks but I need help in trying to get my game to play for others outside of college so that I can get feed back.
    I have three mates who are willing to test it but they don't want to travel into college.
    The day has been proper stressful as I have spent a lot of it disagreeing with my tutor as he was trying to get me to do something that was not in the brief. He really does seem to have something against me. There are only three girls on the course including me its mostly males. I am determined to prove him wrong and I am def leaving and going to another college another year with this twat will just p..s me off.

    Help!
     
  15. Joe-Censored

    Joe-Censored

    Joined:
    Mar 26, 2013
    Posts:
    11,847
    As a QA tester by day I can tell you it is not possible to develop and run a regression suite which covers every bit of functionality of the software being tested for something as large as Unity. They likely have regression scripts for ToggleGroup, but didn't include disabling/enabling the GameObject which it sits on during the test.

    QA when developing test suites tries to target as much functionality as possible, which impacts as many users as possible, within the resources and time allowance they have to run the tests. The fact that this issue appeared in 2019.1.0a5, but no one reported it throughout the rest of the alpha nor the beta, tells us either the ToggleGroup isn't that commonly used, or disabling/enabling its GameObject is a rather obscure use case for it. I'm not surprised when QA misses things like that. They don't have infinite CPU cycles, infinite time to write and maintain test scripts, nor infinite creativity to come up with ideas on every possible way a component can fail.
     
    Peter77 and Lurking-Ninja like this.
  16. AndersMalmgren

    AndersMalmgren

    Joined:
    Aug 31, 2014
    Posts:
    5,358
    It feels like a pretty basic test to disable the parent, something that should be done for all automated uGUI component tests even. But sure maybe they missed that one, but its not the first time simple features like these break. I hope atleast they do a TDD red green approuch when fixing it, so it cant happen again.

    Writing good automated test that test relevant things are a thing of art, most dont master it sadly.
     
    Last edited: Jul 11, 2019
    Joe-Censored likes this.
  17. Joe-Censored

    Joe-Censored

    Joined:
    Mar 26, 2013
    Posts:
    11,847
    All true. The hardest part in QA is writing tests of a feature where it interacts with other features. It is pretty simple to test ToggleGroup's functionality for example (not really that complicated of a feature), but interacting with other engine functionality the possibilities are endless. What happens if I disable/enable the canvas? What happens if I change resolution? What happens if I destroy the object on the same frame a toggle is clicked? What happens if some of the toggles are on windows which are currently disabled? What happens if the GC cleans up on the same frame a toggle is clicked? What happens if the frame rate plunges on the same frame something else in togglegroup changes? What happens.... etc etc.

    Do the same with every feature and how it interacts with every feature, using every option of every feature, and your test cases will be in the millions. Then run those millions against every supported Editor OS, under every language setting, compiling to every supported platform, for every Unity release, and it is just impossible. You can't script that many tests, you can't run that many even if they were scripted already, you can't investigate that many results, and still ship Unity releases on any reasonable time frame. QA has to cut down what could be years of testing into what they can fit into a couple weeks, and stuff gets missed.

    The biggest failure I see here specifically is whoever was verifying whatever change went into 2019.1.0a5 that broke this, didn't understand the change enough or was not thorough enough to check for this problem. That's where the QA person needs to get creative and think outside the box on their own on what this code change potentially touches and test it manually, even though what they test probably won't end up in a regression suite.
     
    AndersMalmgren likes this.
  18. Vryken

    Vryken

    Joined:
    Jan 23, 2018
    Posts:
    2,106
    I'd report this person. Clearly they care more about "being right" than educating.
     
  19. RecursiveFrog

    RecursiveFrog

    Joined:
    Mar 7, 2011
    Posts:
    350
    Agree, from the perspective of a professional in the industry, I find the requirement to use an unstable bleeding edge alpha release to be questionable, on many levels.

    On the bright side, OP you can take this as a lesson about the dangers of using half baked software. I would watch out though, because any lesson plans written based around this version could be a bit... broken
     
  20. AndersMalmgren

    AndersMalmgren

    Joined:
    Aug 31, 2014
    Posts:
    5,358
    Yeah it's a common problem that programmers don't really care what the side effects of a change are. I often see my programmers use existing methods without checking how they are implemented, which often is fine. But in a thight loop etc it can be relevant. Or sometimes they even change a method without checking all it's usages and make sure there are unit tests covering all those usages.

    The last line of defense are you testers, we programmers should make sure as much of the code have automated test coverage. At my day job the system I'm in charge of have some 90 plus procent coverage you can be pretty confident doing refactoring etc in that system, always need to think about what you do offcourse. Our sister system is an old .net system from around 2005 with some questionable aritechture and low code coverage, the poor testers have a hard time with regression on that system :)
     
    Joe-Censored likes this.