Add RFC for shorter vector constructor

[petrihakkinen]

Add RFC for replacing the current vector constructor (vector.create) with a less verbose version (vector).

Rendered

[petrihakkinen]

After a long think I decided to propose the deprecation and eventual removal of vector.create in favor of vector in the RFC because I think it would be better for the language in the long run to have only one constructor. If you think this is a too big breaking change we can modify the RFC to keep the old version as an alias. I don't have a strong opinion on that but we'd prefer some officially sanctioned way to disable vector.create (or any other built-in) locally for our codebase in case vector is added but vector.create is not deprecated.

[jackdotink]

Personally I would like to see special syntax for creating vectors, such as <x, y, z>.

[vegorov-rbx]

I don't think we will be adopting this.
Reasoning for current constructor was already discussed in the vector library RFC and we have selected that option to go forward.

In a way, this RFC was already rejected in the discussion of that RFC.

[petrihakkinen]

I read through the previous discussions again and I guess the main reason for choosing vector.create is consistency with buffer.create? That is a valid point but at least in our codebase the number of vector creations far outweigh buffer creations. For example, in one of our products there is over 15000 calls to the vector constructor where as buffer.create calls can be counted with fingers. Many of the calls are inline as function arguments (colors, offsets and the like) where extra verbosity starts to affect readability. Granted, buffers is a newer feature but vector is such a commonly used primitive in the domains of games and graphics that we think it could justify having a bit shorter syntax for them.

[alexmccord]

I don't think we will be adopting this. Reasoning for current constructor was already discussed in the vector library RFC and we have selected that option to go forward.

In a way, this RFC was already rejected in the discussion of that RFC.

I don't think this accurately conveys what actually happened with the syntax for vector creation. We talked about a whole bunch of them:

1. vector(x, y, z)
2. [x, y, z] following usual notation in linalg
3. <x, y, z>
4. |x, y, z|

I feel a strong urge to mention that I never liked the last two, but I digress.

The reason why we split out the vector creation shorthand from the RFC was to allow it to go through, and then let someone talk about the shorthand separately if they're annoyed by vector.create(x, y, z) being the way to create them.

So I don't think it's fair to automatically reject this RFC.

[vegorov-rbx]

I did not mean to reject it automatically, I reject it based on what is proposed.

[bradsharp]

Also want to mention table.create as additional art.