Ping @alainfrisch on this old one

See also #2236 which proposes to add an of_list with exactly the same question.

(Personally as I mentioned I think that the {add,replace}_{seq,list} are better building blocks and that we should de-emphasize of_* functions that force us to duplicate the API of the create all over the place. Unfortunately this tells nothing about what we should do with the existing of_* functions: leave them frozen in the errors of their ways, or bloat the API just for them?)

I agree, @gasche. The problem with the approach I put here is that this breaks:

seq |> Hashtbl.of_seq

I think it's better to document that of_seq always uses a "default" interpretation of random Hashtbl of size 16 and note to use add_seq if more control is needed:

(* with this PR:
seq |> Hashtbl.of_seq ~size:32 ~random:true
*)
seq |> Hashtbl.add_seq (Hashtbl.create ~random:true 32)

I have one question before closing this. Given that #2236 is never going to land, should we:

1. Just update the documentation for Hashtbl.of_seq as above
2. Deprecate it in favour of always writing the version with add_seq?

There is a consensus against the change proposed in this PR, so closing.

I think it's better to document that of_seq always uses a "default" interpretation of random Hashtbl of size 16 and note to use add_seq if more control is needed

I agree. (I would document this, but not deprecate the function). Anyone, please feel free to open a PR with this documentation improvement.