Helper function on SetHashes

[lizmat]

See rakudo/rakudo#2942

Perhaps similar helper functions could be useful on BagHashes (and MixHashes?) as well. But probably should then be called .add and .subtract

[AlexDaniel]

%sethash{$_} = True for @to-be-added;
%sethash.set(@to-be-added);

I don't think we should be adding stuff purely for speed reasons when other ways of doing the same thing are not ironed out in terms of performance.

For example, I can see other ways of doing the same thing. Like:

%sethash{@to-be-added} X= True;

Which is currently much slower, but there has to be a way to make it as fast as the proposed .set() method, right?

Also, it seems that good performance is already achievable with something like this:

%sethash ,= SetHash.new(@to-be-added);

Although I'm not sure if it is fully identical to the proposed .set method, and if it is as fast?

Edit: more stuff

[lizmat]

my @a = ^10; my %s is SetHash; %s ,= SetHash.new(@a); dd %s.elems   # 2

That should have been 10. You get a SetHash with itself, and the new SetHash. Which, incidentally, infiniloops on .gist or .perl :-(

[lizmat]

Some benchmarks / readability comparison:

$ 6 'my @a = ^100000; { my %s is SetHash; %s{$_} = True for @a; say now - ENTER now }'
0.5418981

$ 6 'my @a = ^100000; { my %s is SetHash; %s{@a} = True xx *; say now - ENTER now }'
0.5311904

$ 6 'my @a = ^100000; { my %s is SetHash; %s.set(@a); say now - ENTER now }'
0.11501774

[vrurg]

@lizmat what is the current status of this? Is it still planned?

[lizmat]

I guess either it is up to @jnthn to decide, or we need to take a vote on it?

The conflicts should be easy to resolve.

[vrurg]

That's a question. We can't rely upon Jonathan on each end every case. As it was said, he is a scarce resource. But we probably need a collective decision making procedure. I'd try to think out something on this.

[jnthn]

We can't rely upon Jonathan on each end every case

No, but language design things probably are one of the things I can most usefully opine on. Design by what can win a vote isn't going to produce much consistency.

Some questions:

1. Were push and append considered? For example, we have Hash.push and Hash.append, which seem to be in this very space, but not on SetHash etc.
2. I wonder if we're missing a "single argument rule" style thing here, so that ∪= could potentially do the job.