Unity DEVs and/or people that just know. I’ve read the blog (http://blogs.unity3d.com/) and I see there are some major improvements and changes coming our way. Is there anything we should really not use or be aware of to avoid having to redo a ton of work?
I would avoid everything and enjoy a nice cup of lemonade while you wait for the new version to come out… just kidding of course.
You should be able to use everything that you usually do without any worry. Here are the things that (according to the roadmap) will be rewritten so run a greater risk of breaking that other areas.
You’re prefabs might brake when they upgrade to 3.5 because they are completely rewriting the system, but knowing the guys at Unity, they will probably repackage your prefabs for you when they make a new release. Just like scripts loaded into 3.x from 2.x were automatically marked up at the top with
They expect to create a new GUI system for 3.5. I doubt the will convert your old gui into the new gui, but at the same time, I expect that the old gui will still work. GUITextures were not removed when the procedural
OnGUI() was created and I expect something similar with this new interface.
Occlusion culling is getting an upgrade, but I doubt that will involve anything more than re-baking your oc to fix.
A new particle system in 3.5. I don’t really know about this one. It will be curve driven so it could be broken, or easily compatible.
Having said that, I doubt much (if anything) will break with 3.4 / 3.5 at least as a known consequence. There will obviously be some bugs introduced with any new version, but I truly doubt that the Unity team will break some components especially for a 3.x release. Come 4.0 I wouldn’t be surpised if something breaks, but not for 3.4 / 3.5.
Note: this is all speculation and should be treated as such. I wish I could predict the future, but in my current skill set, I cannot.
There is generally very little breakage with Unity .x versions. You shouldn’t avoid anything.
If anything, you may end up ditching things you’ve already written in favor of a native solution (if it’s halfway decent) such as pathfinding in 3.5, or light probes.