More conversions

[sebcrozet]

Added

• Conversion from an array [T; D] to an isometry Isometry<T, _, D> (as a translation).
• Conversion from a static vector SVector<T; D> to an isometry Isometry<T, _, D> (as a translation).
• Conversion from a point Point<T; D> to an isometry Isometry<T, _, D> (as a translation).
• Conversion of an array [T; D] from/to a translation Translation<T, D>.
• Conversion of a point Point<T, D> to a translation Translation<T, D>.
• Conversion of the tuple of glam types (Vec3, Quat) from/to an Isometry2 or Isometry3.
• Conversion of a glam type Vec2/3/4 from/to a Translation2/3/4.

[Andlon]

I feel that perhaps some of these conversions are a bit too "aggressive"? For example, From<[[T; R]; C]> for SMatrix might cause a lot of confusion about the fact that the nested array

[[1, 2],
 [3, 4]]

is in fact not the matrix

[ 1, 2 ]
[ 3, 4 ]

but rather its transpose. Similarly, is it completely obvious that a Point is directly transformed into an Isometry? An Isometry is documented in terms of a translation, so I suppose implementing From for a vector makes sense. But a point is not a translation, it's a position. Sure, it may be interpreted as such, but doesn't that defeat the purpose of having a semantic distinction between Point and Vector in the first place?

[sebcrozet]

For example, From<[[T; R]; C]> for SMatrix might cause a lot of confusion about the fact that the nested array.

We also have From<[[T; R]; C]> implementation for SMatrix. So if there is an ambiguity, the compiler should catch it and generate an error requiring explicit type annotation. (EDIT: sorry, I misread your comment, see my other response)

Similarly, is it completely obvious that a Point is directly transformed into an Isometry? An Isometry is documented in terms of a translation, so I suppose implementing From for a vector makes sense. But a point is not a translation, it's a position. Sure, it may be interpreted as such, but doesn't that defeat the purpose of having a semantic distinction between Point and Vector in the first place?

I guess this point conversion may sound odd. However, isometries are very often used to represent a position (especially throughout our own code-bases). So I think this makes the Points -> Isometry relevant.

[sebcrozet]

Sorry, I misread your comment about From<[[T; R]; C]>. Yeah, I understand this can cause a confusion. However it appears that this conversion already existed before (see the impl_from_into_asref_2D macro partially removed by this PR); we are just making its implementation nicer here.

[Andlon]

Similarly, is it completely obvious that a Point is directly transformed into an Isometry? An Isometry is documented in terms of a translation, so I suppose implementing From for a vector makes sense. But a point is not a translation, it's a position. Sure, it may be interpreted as such, but doesn't that defeat the purpose of having a semantic distinction between Point and Vector in the first place?

I guess this point conversion may sound odd. However, isometries are very often used to represent a position (especially throughout our own code-bases). So I think this makes the Points -> Isometry relevant.

Fair enough, I don't use isometries that often.

Sorry, I misread your comment about From<[[T; R]; C]>. Yeah, I understand this can cause a confusion. However it appears that this conversion already existed before (see the impl_from_into_asref_2D macro partially removed by this PR); we are just making its implementation nicer here.

Ah. Hmm, well, in that case I would even suggest to remove it altogether, in favor of a method that makes the storage order explicit. I.e. from_col_major_array or similar.

[sebcrozet]

Ah. Hmm, well, in that case I would even suggest to remove it altogether, in favor of a method that makes the storage order explicit. I.e. from_col_major_array or similar.

Yeah, that would definitely make sense (and would allow us to mark this from_col_major_array as const).

[Andlon]

As a final comment, upon further reflection I think a better name would even be from_col_major_nested_array, because from_col_major_array could also indicate that the data is laid out in a flat array.