Sets coverage.

[Ismael-VC]

Add test for Base.rehash!(s::Set{T}).

[Ismael-VC]

Ref: julia-dev

[JeffBezanson]

Thanks for helping with test coverage!

Honestly I'm not sure why there is a rehash! method for Set. Maybe we should just remove it?

[Ismael-VC]

You are welcome!

I've been reading about hashing and it's use in associative arrays, etc. But honestly I don't know if
Base.rehash!(s::Set{T}) is really useful or not for sets.

I'm just trying to change those red lines to green at coveralls. 😄

[Ismael-VC]

@JeffBezanson what I understood was that rehash! is issued every time a Set grows or shrinks, but since a Dict is used internally, it seems to me that that would call rehash!(d::Dict{K, V}, newsz).

The cpp reference seems to also have something like this defined for sets. Perhaps there is indeed a use case? But does it apply to Julia also?

[Ismael-VC]

Honestly I'm not sure why there is a rehash! method for Set. Maybe we should just remove it?

@stevengj could you tell us why do we need Base.rehash!(s::Set{T})?

[StefanKarpinski]

No idea. Did I introduce it?

[StefanKarpinski]

Oh, wrong ping.

[yuyichao]

@StefanKarpinski To be fair, it was @ing you in the original version.

[Ismael-VC]

It was my mistake, I looked at git blame, and misread by one line the name of the author:

• https://github.com/JuliaLang/julia/blame/master/base/set.jl#L39

Which is why I changed it back. So it seems Steven never got the ping.

@stevengj could you please give us some insight on this method when yo have the chance?

He only changed the name from rehash to rehash! though, I'll try to track down who introduced it. But if no one knows why it's there, then I second Jeff about removing it.

I'm sorry for the noise.

cheers

[yuyichao]

@Ismael-VC That is actually still not the right commit this e8d47e4 is the commit it was introduced and it was by @JeffBezanson .

[yuyichao]

Sorry I was also wrong. It was e7ce4cb by @timholy

[Ismael-VC]

Add convenience calls for rehashing

This is relevant particularly to precompilation, which requires
that any module global dicts get rehashed by the __init__
function.

I think this still stands then.

[stevengj]

Yes, it was for precompilation.

[Ismael-VC]

Ok! then, is the test good as it is?

[stevengj]

I suppose so; I guess there's no way to test rehash! without digging into the Set internals.

[StefanKarpinski]

Bump. Should we go ahead with this?

[tkelman]

Looks like this introduced a non-deterministic failure, see #11890. Happens sometimes, so not a big deal, we'll work out what's best to do here while hopefully keeping this method covered for correctness if we can.

[tkelman]

should be fixed by #11892