shoud I upgrade my project from 2017.3.1 to 2019 ?

hello,shoud I upgrade my project from 2017.3.1 to 2019 ?
What are the benefits？

 

What are you hoping to get out of an upgrade?

 

Follow the trend

 

Following the trend is not worth it.

Normal reasons for upgrades are:
1. Some that you experience are fixed.
2. Some feature you can't live without is introduced.

If neither of those things happened, there's no reason to upgrade.

 

There can be a point of having a migration branch thats on latest tech release even if you dont have a need for the features right now. Because its much harder to upgrade in one large move from 2017 to 2019 than going in baby steps. Also when having this branch you always know the effort of moving even if you havent done the actual move. Thats how I do it, when a new major release comes out I just create a new branch and load the project to see what which compile errors we get, which automated tests break.

And one day you might get a feature you want, or Unity drops support for a platform on your old version, etc, etc Than you are in a bad position if you are on a very old version.

edit: Btw, scene loading times are alot faster in 2019.x than in 2017.4.x. Our bigger scenes load much faster. Only real benefit I can come up with if you dont use any of the new features

edit: Update to 2017.4.40 first. Test that. Commit. Than upgrade to 2019.4.15

 

If you want to upgrade, start by making a full backup of your project. Then update to 2017.4. Then update to 2018.4. And then update to 2019.3 (2019.4 when it is available). There might be some upgrade scripts that expect 2017->2018-2019, so you probably want to upgrade that way to catch any of those updates.

 

Ťhanks, As a coder, I feel like I'm being eliminated by frequent updates from unity

 

Dont backup. Use versioning

edit: btw, you might want to go 2017.4.x > 2018.4.x > 2019.3.x

 

I upgraded from 2017.2 to 2019.3 a few months; I had to fix a few obsolete functions in scripts and 2 prefabs that reset and other then that it went smoothly. Remember to backup your entire project (except the Library folder, it takes a very long time to copy and will be regenerated if needed) first.

If you don't need the new features I wouldn't suggest upgrading (ex: if you had 2018.4 it wouldn't be a big deal) but there's a lot of reasons to upgrade from 2017 to 2019, such as: Scene view camera settings (so you can change the max view distance from ∞ to something reasonable), way faster compile times, support for new assets from the Asset Store, and a few other things (like being able to turn off casting and receiving shadows for specific prefabs).

 

There can be at least one more very good reason:
You are using a framework built over Unity, which you are using to create your game and all your future games.
In which case, you have to keep that framework up-to-date or you'll still be using Unity 2017 in 2027.

In fact, I think that there may be a lot of reasons to do that, depending on the exact situation of the dev team.


Anyway, to answer the initial question, upgrading from 2017 to 2019 (or 2020 for that matter) is probably very easy if the project is simple (not using a lot of different Unity features), and probably very hard if the project is complex (using a lot of different Unity features).

So, I'd say that it depends.
I think the best way is to try it for a day or two, and see how it goes.
If it doesn't go well, you can try to upgrade 2017 -> 2018 -> 2019 -> 2020 as already stated.
And anyway, as already stated, do not backup before doing that, use versioning if it's not already the case.


And if for some reason you want to keep up-to-date with Unity in the future, I'd recommend to update to the new Unity version every 2 Unity versions (2020.1 -> 2020.3 -> 2021.1 -> 2021.3 ...), so that each update is only a few hours and any regression can be easily pin-pointed as there aren't a lot of diff with the previous version.

 

I am currently on 2018.2.21f1. Would I notice improved scene loading and compile time performance if I upgraded to 2019.x?

Also, another thing to keep in mind is what assets from the Asset Store you are using. If you have a lot that have not been upgraded to your planned upgrade version, it probably wouldn't be worth it until the assets are updated.

 

yes it is signifcantly faster. but u should only upgrade if your work requires the tools available. if you going for simple style 2018 or even lower will be fine.

 

my project includes steamvr and vrtk and hi5 glove sdk.
Full of fear of two things：
1 May cause irreparable errors
2 Unable to adapt to the new editor for a short time （Because unity have a lot of updates）

 

1. Is a non-issue, since you can always use your version control to revert anything that breaks. You are using version control, right?

2. The editor has not changed that much. It won't take much time to adapt.

 

Use version control and revert if upgrade fails.

 

Unity should take upon them self and learn people to use version control. For example when you create a new project it should tell you about it. When you upgrade project they should change that backup message to tell them they need to use version control and commit any changes that they do not want to lose.

Also they should bomb users with info about version control when ever they can. It seems their customers are unaware.

 

Studying version control is user's responsibility.

They shouldn't because in situation where I KNOW that version control is not needed I don't want to be pestered by pointless messages.

Basically, not a good idea to pester people who know what they're doing based on mistakes of those who are unaware. Because this is made for people who know what they're doing.

 

Then they should remove the message box thats present today when upgrading with your mindset. Dont forget that 99% of unity user base are noobs

edit: there is almost no situation when vcersion control is not needed. Even POC projects benefit from it.

only time I dont use it is repro projects

 

This requires proof, otherwise you're throwing random insults at userbase.

Additionally the engine is made for skilled users who know what they're doing, because those are the ones bringing in the money to the company. They're the target audience, despite lowered barrier of entry.

A quick 15 minute test does not require version control.

 

Until you expand it for more tests, and at some point something is broken and you don't know what.

On the other hand, how much time does it take to create a SVN/GIT repo on your HDD and to start working with it?
2-3mn, no more, and even less if you've already done it often enough.


That said, I admit that I am sometimes (often... very often...) too lazy to do even that, especially for 15mn tests as you said.
Note that twice I had the problem I underlined, so now I work on a system which saves all the modifications and allow me to get back to any previous state if I want to. It's not as great as SVN or GIT, but for the rare cases where a small test is causing me some trouble it can save the day.

 

Another good reason is if you're targeting a platform with a small(ish) version compatibility window. For instance, if I were targeting mobile I'd want to keep my project reasonably up to date because every so often the device vendors change stuff that I need Unity to support.

I don't follow the logic, here. Whether you spread out the steps over time or take them all one after the other the effort required to upgrade is the same.

Just because Unity have released multiple updates doesn't mean you have to step to the latest directly from whatever you're using. I'd still go one moderate update at a time. But any effort I spend doing that before I know there's a benefit is effort I could have put into something else if it turns out I never need it.

At which point it's no longer a "15 minute test" and yeah, you should have put it in VC at the point where it became important.

In any case, I agree it's not on Unity to forcefully educate people on that. Experienced people should know on their own. Newbies have enough on their plate already, and there's a bunch of foundational knowledge people need to have before version control is helpful. Try to start doing that stuff too early and it'll hinder rather than help. You can't learn all of the things first.

Given the fact that Unity is deliberately accessible to newbies I think it would be a really good idea for them to start and maintain a knowledge base of general game dev info, and version control would be a part of that. But the tools themselves shouldn't try to enforce any such thing. I think that reminding people to back stuff up is a pretty good position to take.

 

Do I need to explain the obvious to you?

That is no longer a 15 minutes test.

Programmer should apply practices due to a good reason, and not follow a cargo cult. Don't turn version control into a cargo cult.

Also, git repository does not take 3 minutes to setup. It takes a second. However, for 15 minute test you don't need it.

Do you know when the first time is to implement version control? It is when you start doing things that might BREAK your existing code. Getting to that point takes from few hours to a few days. Depending on what you're doing.

 

This! <3

 

If your project works fine in 2017 and if you dont have a any plans about using hdrp or urp in your project you dont have to upgrade

 

What platform(s) is your project for?