On 2/28/20 7:57 PM, aliak wrote:
> 
> I actually didn't realize it was a video, thought it was just an article! - But anyway, it was just to point out that swift lowers to specialized types when it comes to interpolation (which is what you and adam are trying to get through). And therefor you can detect interpolations being given to you and deal with them the way you want and you can do a lot when you know you're getting an interpolation. You can create types like
> 
> let example: SQLStatment = "select * from blah where a=\(a), b=\(b) ... "

I didn't get to this part of the video, but that is indeed pretty cool. I'm assuming that this generates placeholders for the SQL statement and attaches a and b as parameters?

However, D cannot do something like this exactly, because expressions define the tuple, not how they are used.

But this is possible (with the proposed DIP or ours):

alias sql = "select * from blah where a=$a, b=$b"; // aliased to the tuple

connection.query(sql);
a = 5;
connection.query(sql); // another query with `a` set to 5 now.

Swift can do some pretty cool things due to the type resolver, but it comes at a cost (some expressions that are trivial in D make the compiler complain about them taking too long to resolve).

> I also didn't realize the takeaway would be that swift does appending 😆- which by the way, is not completely accurate. And it does not generate temporaries (unless you mean passing in parameters? There's no way around that if you want to end up with a string based on runtime values - it'll have to be processed in to a string somewhere).

For example, the part where they change the date formatting, they use a date formatter to generate a string for the date, which then is appended to the string interpolation.

Yes, you need to allocate a string. But you should only allocate one.

> You can also get an interpolated string directly in to "print processing" if you wanted to: https://swift.godbolt.org/z/muAzgm

Hm... I'm not too impressed with this when compared to writefln(i"hello $("hello"), $x"); which works without such extra mechanics or strange call syntax.

> 
> When it comes to printing it really doesn't matter if you construct a string on the stack and pass it along. You're IO bound anyway.

IO is not usually processed directly to the device. Usually it's buffered. Writing directly to the buffer is preferable to generating a string on the stack, and just copying that to the buffer.

> One very interesting thing of note is the way they combine named arguments with string interpolations.

Yeah, that part is cool. But that comes for free with Swift. Potentially with the right additions to the named parameters DIP, it would be feasible to do something similar with this DIP.

> Also another note, this tuple expansion should really not be called string interpolation, since it does not result in a string :/ It's more string expansion really.

I recommended calling it a "formatted tuple" to avoid conflation with existing string interpolation implementations.

-Steve

On Saturday, 29 February 2020 at 14:41:16 UTC, Steven Schveighoffer wrote:
> On 2/28/20 7:57 PM, aliak wrote:
>> 
>> I actually didn't realize it was a video, thought it was just an article! - But anyway, it was just to point out that swift lowers to specialized types when it comes to interpolation (which is what you and adam are trying to get through). And therefor you can detect interpolations being given to you and deal with them the way you want and you can do a lot when you know you're getting an interpolation. You can create types like
>> 
>> let example: SQLStatment = "select * from blah where a=\(a), b=\(b) ... "
>
> I didn't get to this part of the video, but that is indeed pretty cool. I'm assuming that this generates placeholders for the SQL statement and attaches a and b as parameters?

I'm not sure if it's in the video (I haven't seen the video). But it was just an example, implementations would vary I'd assume. Here's one from a quick search for e.g.: https://github.com/groue/GRDB.swift/blob/4c8574aa4c08ff715ce0e63b38502ac1e8529069/GRDB/Core/SQLInterpolation.swift

>
> However, D cannot do something like this exactly, because expressions define the tuple, not how they are used.

Yep, and similarly expressions define the interpolation pattern in swift, not how they are used as well. So the tuple thing is also pretty sweet! I'd imaging you'd be able to do something like:

auto sql = SQLStatment(i"select * from blah where a=$a, b=$b");

where

struct SQLStatement {
  string formatted;
  // What template magic is needed here? This is where I get a little worried.
  this(S : _interpolated_object!Specs, Specs, Args...)(S s, Args args) {
    // build formatted...?
  }
  alias formatter this;
}

>
> But this is possible (with the proposed DIP or ours):
>
> alias sql = "select * from blah where a=$a, b=$b"; // aliased to the tuple
>
> connection.query(sql);
> a = 5;
> connection.query(sql); // another query with `a` set to 5 now.

Yeah, and that's pretty sweet too!

>
> Swift can do some pretty cool things due to the type resolver, but it comes at a cost (some expressions that are trivial in D make the compiler complain about them taking too long to resolve).

Yeah, swift ha a few problems with its type solver during semantic I think.

let a: Double = -(1 + 2) + -(3 + 4) + -(5)

Will fail to compile on my machine, and you need to break it up.

But, on the other hand, swift's constraint solver avoids backwards situations like:

void f(bool) { "bool".writeln; }
void f(int) { "int".writeln; }
enum A { one }
void main() {
    f(0);
    f(A.one);
}

And:

void main() {
    struct B {
        B opBinary(string op : "+")(int b) { return this; }
    }
    static if (is(typeof(B.init + size_t.init))) {
        size_t x = 1;
        B b1, b2;
        b1 = b2 + x; // fails here
    }
}


>
>> I also didn't realize the takeaway would be that swift does appending 😆- which by the way, is not completely accurate. And it does not generate temporaries (unless you mean passing in parameters? There's no way around that if you want to end up with a string based on runtime values - it'll have to be processed in to a string somewhere).
>
> For example, the part where they change the date formatting, they use a date formatter to generate a string for the date, which then is appended to the string interpolation.
>
> Yes, you need to allocate a string. But you should only allocate one.

Generating temporaries is not a constraint that the interpolation system puts on you.

>
>> You can also get an interpolated string directly in to "print processing" if you wanted to: https://swift.godbolt.org/z/muAzgm
>
> Hm... I'm not too impressed with this when compared to writefln(i"hello $("hello"), $x"); which works without such extra mechanics or strange call syntax.

Indeed.