[TODO] set inconsistencies

[timotheecour]

• ['a', 'b'].toSet returns a HashSet[char], not a set[char]
• ['a', 'b'].toHashSet is not defined
• card is defined for both set and HashSet, but len is defined for HashSet only, not set
• echo({'a', 'b', 'c', 'b'}.card) : 4 (that one is already reported here: set bug: echo({'a', 'b', 'c', 'b'}.card) : 4 #7555)

[cooldome]

I can agree only with point 3 on len

[narimiran]

Number 4 was resolved a day ago by #7558.
len and card in HashSets are the same function with a different name - number 3 could be easily solved by aliasing.

That leaves first two points to consider/solve.

[timotheecour]

/cc @narimiran

That leaves first two points to consider/solve.

Well assuming number 3 could be easily solved by aliasing is done :-)

actually i don't even know why we need card at all, seems like pointless aliasing that introduces a new symbol ; len is what's used everywhere else. Why not mark card as deprecated?

[krux02]

@timotheecour you could make it a pr that uses the deprecated pragma, I would vote for it.

[metagn]

The deprecated pragma doesn't work for procs.

set and HashSet are different things, set is not a linear type, as in you can't do for i in 0..<x.len: echo x[i] easily. HashSet is a length attribute and a seq of hashes and elements, so it's easy to know that its length is the length of the seq it uses, but set has a different implementation that might make the term "length" misleading.

[andreaferretti]

The length of a set is essentially one or more popcount operations, which are exposed in bitops, so it seems to me not really problematic

[narimiran]

Not an inconsistency between sets and HashSets, but while we're talking about what should be implemented:

The thing missing from both versions is pop, which should remove and return a random element from a (hash)set.

[krux02]

Well I disagre on pop.

[Araq]

I think the rest is fine, at one point you have to stop whining and read some documentation.

[timotheecour]

#11885 fixed item 3 in top post

[timotheecour]

remaining item in this list would be fixed by merging #16276 and removing the (now deprecated) std/sets.toSet.