Thread overview
bolts meta programming library version 1.0.0 - including the from idiom
Jul 15
aliak
Jul 16
Aliak
Jul 17
victoroak
Jul 17
aliak
July 15
I've been using a set of meta tools for a while now, so decided to release it as 1.0.0 with a few enhancements chucked on.

Two of the highlights are the non-eponymous "member" and "iz" templates, which are shown below with some code. The library also includes the "from" template because I tend to use that everywhere.

Docs: https://aliak00.github.io/bolts/bolts.html
Github: https://github.com/aliak00/bolts
Package: https://code.dlang.org/packages/bolts

Current issues include:

1) that I'm not completely sure about the behaviour of the copy-constructor traits - I've decided to pretend post blots don't exist and marked the library as requiring dmd frontend 2.086 and above.

2) It's an ongoing battle with D to normalize parameters because traits return tuples, or string, or tuples of strings, or alias tuples, etc. And some traits take symbols, while others take strings, etc. One of the goals is to make this library a little more consistent and try and keep it that way. I will break things to make things consistent (is my intention).

3) Naming - e.g. why are some things in std.meta capitalized (i.e. Filter) and others camel-cased with a static prefix - i.e. staticMap? What would be the consistent thing to do? I currently have a staticZip in there, should it just be Zip?

Some sample code:

unittest {
    import bolts;

    int i;
    class C { void f0() {} int f1(int) { return 0; } }
    static struct S { void f0() {} int f1(int) { return 0; } @property void set(int) {} }

    pragma(msg, from.std.allSatisfy!(iz!int.of, 3, 4, int, i)); // true
    pragma(msg, from.std.Filter!(isNullable, int*, C, S)); // (int*, C)

    // Member functions
    pragma(msg, memberFunctionsOf!S.asStrings); // tuple("f0", "f1")
    pragma(msg, memberFunctionsOf!C.asAliases); // tuple(f0, f1, toString, toHash, opCmp, opEquals, factory)

    // member template
    pragma(msg, member!(S, "f0").exists); // true
    pragma(msg, member!(S, "f0").protection); // public
    pragma(msg, member!(S, "f0").isProperty); // false
    pragma(msg, member!(S, "set").propertySemantics); // PropertySemantics.w

    // iz template
    static struct OldS {
        this(this) {}
    }
    static struct NewS {
        this(ref NewS s) {}
    }
    pragma(msg, iz!S.triviallyCopyConstructable); // yes
    pragma(msg, iz!OldS.triviallyCopyConstructable); // no because post-blit
    pragma(msg, iz!OldS.copyConstructable); // no because post blit

    pragma(msg, iz!NewS.copyConstructable); // yes
    pragma(msg, iz!NewS.nonTriviallyCopyConstructable); // yes

    // meta fun with packs and zip
    {
        alias a = AliasPack!(1, 2, 3);
        alias b = AliasPack!(4, 5, 6);
        alias c = AliasPack!(7, 8, 9);

        alias d = staticZip!(a, b, c);

        static assert(d.length == 3);

        static assert(d.Unpack[0].equals!(1, 4, 7));
        static assert(d.Unpack[1].equals!(2, 5, 8));
        static assert(d.Unpack[2].equals!(3, 6, 9));

        static assert(AliasPack!(d.UnpackDeep).equals!(1, 4, 7, 2, 5, 8, 3, 6, 9));
    }
}

Cheers,
- Ali

July 15
On Monday, 15 July 2019 at 11:13:10 UTC, aliak wrote:
> I've been using a set of meta tools for a while now, so decided to release it as 1.0.0 with a few enhancements chucked on.
>
> [...]

Nice! I'm working on something similar but with a different goal.
July 16
On Monday, 15 July 2019 at 21:20:16 UTC, Atila Neves wrote:
> On Monday, 15 July 2019 at 11:13:10 UTC, aliak wrote:
>> I've been using a set of meta tools for a while now, so decided to release it as 1.0.0 with a few enhancements chucked on.
>>
>> [...]
>
> Nice! I'm working on something similar but with a different goal.

Thanks! How’s the thing similar and what’s the goal (if you don’t mind me asking)?
July 16
On Tuesday, 16 July 2019 at 00:10:19 UTC, Aliak wrote:
> On Monday, 15 July 2019 at 21:20:16 UTC, Atila Neves wrote:
>> On Monday, 15 July 2019 at 11:13:10 UTC, aliak wrote:
>>> I've been using a set of meta tools for a while now, so decided to release it as 1.0.0 with a few enhancements chucked on.
>>>
>>> [...]
>>
>> Nice! I'm working on something similar but with a different goal.
>
> Thanks! How’s the thing similar and what’s the goal (if you don’t mind me asking)?

Similar: One-stop shop for reflection unifying the disparate APIs.
Goal: allow "regular" code for reflection purposes by (also) returning everything as strings.
July 16
On Tuesday, 16 July 2019 at 18:18:50 UTC, Atila Neves wrote:
> On Tuesday, 16 July 2019 at 00:10:19 UTC, Aliak wrote:
>> On Monday, 15 July 2019 at 21:20:16 UTC, Atila Neves wrote:
>>> On Monday, 15 July 2019 at 11:13:10 UTC, aliak wrote:
>>>> I've been using a set of meta tools for a while now, so decided to release it as 1.0.0 with a few enhancements chucked on.
>>>>
>>>> [...]
>>>
>>> Nice! I'm working on something similar but with a different goal.
>>
>> Thanks! How’s the thing similar and what’s the goal (if you don’t mind me asking)?
>
> Similar: One-stop shop for reflection unifying the disparate APIs.
> Goal: allow "regular" code for reflection purposes by (also) returning everything as strings.

Is this related to the talk Andrei gave at a recent dconf?
July 17
On Monday, 15 July 2019 at 11:13:10 UTC, aliak wrote:
> I've been using a set of meta tools for a while now, so decided to release it as 1.0.0 with a few enhancements chucked on.
>
> [...]

Looks nice. Though, I see it has some problems that stuck me before. The traits isFunctionOver and isUnaryOver can't handle lambdas that receive arguments by ref like:

struct S {}
static assert(isFunctionOver!((ref s) => s, S));

I wish ref was part of the type in D, so we could pass ref S as template parameters. I could pass as string but it's error prone and not really elegant.
July 17
On Wednesday, 17 July 2019 at 16:29:34 UTC, victoroak wrote:
> On Monday, 15 July 2019 at 11:13:10 UTC, aliak wrote:
>> I've been using a set of meta tools for a while now, so decided to release it as 1.0.0 with a few enhancements chucked on.
>>
>> [...]
>
> Looks nice. Though, I see it has some problems that stuck me before. The traits isFunctionOver and isUnaryOver can't handle lambdas that receive arguments by ref like:
>
> struct S {}
> static assert(isFunctionOver!((ref s) => s, S));
>
> I wish ref was part of the type in D, so we could pass ref S as template parameters. I could pass as string but it's error prone and not really elegant.

Ooh! Thanks!

Fixed and released v1.0.1
July 23
On Tuesday, 16 July 2019 at 20:18:30 UTC, John Colvin wrote:
> On Tuesday, 16 July 2019 at 18:18:50 UTC, Atila Neves wrote:
>> On Tuesday, 16 July 2019 at 00:10:19 UTC, Aliak wrote:
>>> On Monday, 15 July 2019 at 21:20:16 UTC, Atila Neves wrote:
>>>> On Monday, 15 July 2019 at 11:13:10 UTC, aliak wrote:
>>>>> [...]
>>>>
>>>> Nice! I'm working on something similar but with a different goal.
>>>
>>> Thanks! How’s the thing similar and what’s the goal (if you don’t mind me asking)?
>>
>> Similar: One-stop shop for reflection unifying the disparate APIs.
>> Goal: allow "regular" code for reflection purposes by (also) returning everything as strings.
>
> Is this related to the talk Andrei gave at a recent dconf?

Yep!