Should I update my project or keep using the version I started with?

Do people continue to update their game as new stable updates are released or do you start a game on one version and release it using that same version?

I feel I run into a lot of problems updating so would think personally I should never do it. Is that the main idea anyway or should I suck it up and stay up to date?

 

Fork your project and test a version of Unity that supposedly fixes your problem; if it works you're good, if it doesn't then explore other options.

 

There are two schools of thought on this.

I'm very firmly in the only upgrade if you need to camp. Upgrading takes effort and testing, and always has an element of risk. I'm not going to spend that effort unless there's some benefit that clearly outweighs the cost. If everything works and there's nothing new I need... why would I upgrade? It's just a bunch of work to change a number.

Note that even with this thinking there are times when you just have to do it. Particularly back when I was doing stuff for mobile, there was only so much of a window where everything was supported at once. It was a pretty long window, but not endless.

If you're taking this approach then I recommend at least checking out the LTS streams. They're not perfect, but nothing is. Basically, they give you bug fix updates without any feature changes, for a period of 2 years.


The other camp is always keep up to date. Depending on what you're making I see where these people are coming from. For instance, if you're on a platform with a limited support window then this could reduce risk, by letting you know ASAP if your specific project has an issue with a specific version of Unity. Knowing about it before you have to upgrade means you've got extra time to deal with it.

If you're using lots of automated testing that can make this approach much easier, too.

 

Theories are nice, but the reality is that support for platform requirement X is only available on version Unity version Y, and version Y, which is an LTS, just broke controller support, so you're F***ed.

Theoretical approaches are only useful if you are casually developing for PC. If it's anything beyond that, there is no saving you.

 

If you know you will not need to update you can stay on the same version. But in reality things change, down the road unity might introduce new features that will improve your game etc. Or there is the possibility like mentioned that you need to upgrade for compability reasons. For example you might need to update third party SDKs to support new hardware, one such real world exam on is steamvr, to support Valve Index fully you need to use steamvr 2.x and they do not support too old versions of unity.

In my opinion its always better to stay ahead and have a active strategy for migrating. We have 3 branches one for production were we only fix bugs. Than we have the dev version were we develop features for upcoming releases. Lastly we have the migration branch we're we use latest version of unity stable (no beta or alpha) and when we have a working version we pull it into dev branch.

 

If the size of the game isn't that large then it wouldn't be much to fork the project and test accordingly. Otherwise I wouldn't recommend updating all that often, especially if you're on an LTS version. Of course, if you're developing for mobile, then there's much less choice there.

 

Not saying anything new here, but if there is decision to be made what you want to do is turn as much unknown into known as you can. Then you can weigh pros and cons for your specific needs.

So if you don't have a workflow that allows non-destructive testing, you should practice that and then integrate into your pipeline (what people already said.)

You can probably do fine just making games in whatever version you want and never worrying about upgrades too. I guess it just depends what kind of games you are making really. As always the question is comes down to time management.