Clarifications on ragged literals

[novo52]

https://discord.com/channels/1392396664299458601/1446720730166202420

This discord thread wins the most confused and frustrated students award.

How do we resolve this?

The main problem here seems to be that vector literal syntax is the same as array literal syntax, so nested array literals can either be vector<integer[*]>, or integer[*][*]. A ragged literal then means different things depending on the context. Additionally, it is ambiguous whether arrays are implicitly promoted to vectors, so what exactly are the objects on the right-hand side of the examples given by students?

[rcunrau]

The discord thread starts with this: vector<integer> a = [[1,2,3], [1,2,3,4,5], [1,2,3,4]][2];

And all the people who don't know anything agree this is fine. And it goes downhill from there. Do you think this is fine?
To me nothing about this looks remotely legal. A vector of integers is [1,2,3]. The above is a multidimensional array of ragged right vectors, and is not legal given everything we've agreed to. The closest I can think of would be integer[vector<integer[*]>][2], but then the [1,2,3,4,5], and [1,2,3,4] fields would generate size errors.

[rcunrau]

Personally, I think the representation of vector<integer[*]> and integer[*][*] are different, specifically because the former is a vector of arrays and the second really is a matrix. I think that means that they are not compatible as two dimensional objects, but you can still take one-dimensional slices of the vector. I don't think Rust defines 2D slices and we don't either (but if you can find 2D slices in Rust I'd love to see them!). That also answers the questions: there is no promotion/conversion from a multidimensional vector to a multidimensional array. They're just different