On Mon, Jul 18, 2022 at 6:43 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Um ... why is "the order in which the elements were chosen" a concept
> we want to expose? ISTM sample() is a black box in which notionally
> the decisions could all be made at once.
I agree with that. But I also think it's fine for the elements to be
returned in a shuffled order rather than the original order.
> > I really think this function needs to grow an algorithm argument that can
> > be used to specify stuff like ordering, replacement/without-replacement,
> > etc...just some enums separated by commas that can be added to the call.
>
> I think you might run out of gold paint somewhere around here. I'm
> still not totally convinced we should bother with the sample() function
> at all, let alone that it needs algorithm variants. At some point we
> say to the user "here's a PL, write what you want for yourself".
I don't know what gold paint has to do with anything here, but I agree
that David's proposal seems to be moving the goalposts a very long
way.
The thing is, as Martin points out, these functions already exist in a
bunch of other systems. For one example I've used myself, see
https://underscorejs.org/
I probably wouldn't have called a function to put a list into a random
order "shuffle" in a vacuum, but it seems to be common nomenclature
these days. I believe that if you don't make reference to Fisher-Yates
in the documentation, they kick you out of the cool programmers club.
--
Robert Haas
EDB: http://www.enterprisedb.com