Fix issue 9591 - Add ApplyWith template

[schveiguy]

Bikeshedding for ApplyWith name welcome. If you look at the example, it reads well, but maybe not when specified on its own.

[dlang-bot]

Thanks for your pull request, @schveiguy!

Bugzilla references

Auto-close	Bugzilla	Description
✓	9591	std.typetuple.staticApplyMap

[wilzbach]

Spacing after foreach applies to static foreach too (that's why CircleCi is failing)

[schveiguy]

Spacing

I will never be trained at this point to do that spacing properly... :)

Updated.

[edi33416]

I kept looking at this and I might be wrong, but I can't stop wondering if it wouldn't be more natural for the call to be A0!(A1, ..., An) ?
If so, maybe we could have the template declaration as ApplyWith(Func, Args...) ?
What do you think?

[schveiguy]

I get what you are saying. By itself, ApplyWith (or whatever we should call it) looks weird when you add the template at the end. The whole point of this template is to work with ApplyLeft. It's like specifying all the parameters, and leaving the actual template to be used up to a later time.

For example:

template T1(T) {}
template T2(T) {}

alias X = ApplyLeft!(ApplyWith!string);

alias a = X!T1; // equivalent to T1!string
alias b = X!T2; // equivalent to T2!string

I'm thinking now, since pretty much ApplyLeft is required to make any sense, ApplyWith should use it internally instead of Instantiate. The fact that it's used can be hidden.

[schveiguy]

OK, I redesigned ApplyWith so it is much less confusing. Thanks @edi33416 for the review, please take a look at the new design.

[dnadlinger]

Isn't that simply ApplyRight!(Instantiate, Args)?

[dnadlinger]

(And if so, alias functions = staticMap!(ApplyRight!(Instantiate, string), leftJustify, center, rightJustify); imho reads just fine and is self-explanatory, no need for ApplyWith.)

[schveiguy]

@klickverbot Yes, I didn't think of that mechanism. However, Instantiate is private. Maybe I can just close this and make a PR to make it public?

[schveiguy]

Opened said PR: #5739