Luau type syntax now allows for up to 1000 union or intersection type options. Previously, the effective limit was between 250 and 1000, depending on the individual option type.
Wow that’s a huge upgrade! Was there a problem initially preventing this or…?
Luau buffer type and library when?
Changes the type signature of Vec2/3 Min/Max API to a warning when supplying fewer than two arguments.
This is annoying. If I want to create an empty vector, normally I just do Vector3.new(), if this releases I’ll get tons of little type check errors everywhere.
guess we have to use Vector3.zero from now on
edit: maybe not
Vector2.zero and Vector3.zero are your friends!
You asked incorrectly, you should ask “How’s buffer?”
Correct me if I’m wrong but I think this is for vector:Min() and vector:Max() when there’s no second argument, not Vector3.new(). Hope this helps.
Thank god.
The ability to disable force limits for LinearVelocity and the increase of maximum unions to 1000 are definitely the best features in this release note for me.
I can’t seem to find the right post talking about the Min/Max function, but why do they behave in this way where the highest/minimum values get placed into a new vector? There is nowhere in my huge code base where I have ever needed to do such thing, it feels like it can be so random at times unless you know how each vector looks like. If there is a use-case for this, can you guys let me know what it is?
You know what I thought it was? Vector3.new(1, 5, 3):Max() → 5
That’s what I use a lot in my game, to the point that I myself added a function for that, same for minimum.
There are so many features that are getting released under a specific name that makes me think is the solution I have always wanted and are totally what I don’t expect or even need. Like making Constraints have a limit/max force, I expected to place a Vector3.new(1, 0, 1) to let the constraint know where I want to apply force, no the way is working that I never get it right to apply a constant velocity…
Switching to that for an AlignPosition is just like having a BodyPosition…
And so I watch the sun set over a grateful universe.
(It’s technically just swapping the byte order of any integer passed to it, so it’s not just big-endian. Long-term it’ll compile down to a single instruction with native codegen but AFAIK that isn’t implemented yet. It’s still a lot easier than doing it using other methods though, and generally faster.)
When will the Crashing be fixed?
When will you do an update for performance improvements?
When will you continue these questions above?
Release mah boy, he ain’t do nothing. 
Seeing as this is a call inside renderstepped, I’m assuming you’re thinking what I’m thinking - time for fast camera head look directions !!!