Thread overview
how to iterate over an AA by key-value pair (tuple)?
Feb 13, 2014
Timothee Cour
Feb 14, 2014
timotheecour
Feb 14, 2014
bearophile
Feb 14, 2014
Timothee Cour
Feb 14, 2014
bearophile
Feb 14, 2014
Meta
Feb 14, 2014
bearophile
Feb 14, 2014
Jakob Ovrum
February 13, 2014
how to iterate over an AA by key-value pair (tuple)?
use case:
avoid interrupting UFCS chains, eg:
foo.generate_aa.byKeyValue.filter!(a=>a[0].isLower).map!(a=>a[1]) ...
of course I could roll my own function but I was wondering if there's
already a way.


February 14, 2014
On Thursday, 13 February 2014 at 23:56:35 UTC, Timothee Cour
wrote:
> how to iterate over an AA by key-value pair (tuple)?
> use case:
> avoid interrupting UFCS chains, eg:
> foo.generate_aa.byKeyValue.filter!(a=>a[0].isLower).map!(a=>a[1]) ...
> of course I could roll my own function but I was wondering if there's
> already a way.


is there anything more efficient than this?

auto byKeyValue(T)(T a)if(isAssociativeArray!T){
   return a.byKey.map!(b=>tuple(b, a[b]));
}
February 14, 2014
timotheecour:

> is there anything more efficient than this?
>
> auto byKeyValue(T)(T a)if(isAssociativeArray!T){
>    return a.byKey.map!(b=>tuple(b, a[b]));
> }

zipping byKey and byValue could be faster, but there's no guarantee its result is correct.

Bye,
bearophile
February 14, 2014
On Thu, 13 Feb 2014 19:06:45 -0500, timotheecour <timothee.cour2@gmail.com> wrote:

> On Thursday, 13 February 2014 at 23:56:35 UTC, Timothee Cour
> wrote:
>> how to iterate over an AA by key-value pair (tuple)?
>> use case:
>> avoid interrupting UFCS chains, eg:
>> foo.generate_aa.byKeyValue.filter!(a=>a[0].isLower).map!(a=>a[1]) ...
>> of course I could roll my own function but I was wondering if there's
>> already a way.
>
>
> is there anything more efficient than this?
>
> auto byKeyValue(T)(T a)if(isAssociativeArray!T){
>     return a.byKey.map!(b=>tuple(b, a[b]));
> }

Of course. The a[b] lookup is significant in the complexity, it's amortized constant, and the constant is not always small. Consider that all keys and values are already stored in structs inside the AA. With low-level access, it would be trivial to make an efficient tuple generator that did not need to lookup values by keys.

-Steve
February 14, 2014
On Thu, Feb 13, 2014 at 5:50 PM, Steven Schveighoffer <schveiguy@yahoo.com>wrote:

> On Thu, 13 Feb 2014 19:06:45 -0500, timotheecour <timothee.cour2@gmail.com> wrote:
>
>  On Thursday, 13 February 2014 at 23:56:35 UTC, Timothee Cour
>> wrote:
>>
>>> how to iterate over an AA by key-value pair (tuple)?
>>> use case:
>>> avoid interrupting UFCS chains, eg:
>>> foo.generate_aa.byKeyValue.filter!(a=>a[0].isLower).map!(a=>a[1]) ...
>>> of course I could roll my own function but I was wondering if there's
>>> already a way.
>>>
>>
>>
>> is there anything more efficient than this?
>>
>> auto byKeyValue(T)(T a)if(isAssociativeArray!T){
>>     return a.byKey.map!(b=>tuple(b, a[b]));
>> }
>>
>
> Of course. The a[b] lookup is significant in the complexity, it's amortized constant, and the constant is not always small. Consider that all keys and values are already stored in structs inside the AA. With low-level access, it would be trivial to make an efficient tuple generator that did not need to lookup values by keys.
>
> -Steve
>

That's what I was suspecting,
I guess the code to modify would be here:

druntime/import/object.di:455:5

auto byKeyValue(){...}

auto byKey()
    {
        static struct Result
        {
            AARange r;

            @property bool empty() { return _aaRangeEmpty(r); }
            @property ref Key front() { return
*cast(Key*)_aaRangeFrontKey(r); }
            void popFront() { _aaRangePopFront(r); }
            Result save() { return this; }
        }

        return Result(_aaRange(p));
    }

That seems like a worthy enhancement.
If I/someone does it, will it be merged in?


February 14, 2014
Timothee Cour:

> auto byKeyValue(){...}

It's probably better to call the method "byPair" that is shorter.


> If I/someone does it, will it be merged in?

Some people want those pairs to be tuples. But D lacks built-in tuples, so you need to use typecons ones. But to use the typecons ones you need to import half Phobos. It seems there is no good solution.

Bye,
bearophile
February 14, 2014
On Friday, 14 February 2014 at 07:35:34 UTC, Timothee Cour wrote:
> That seems like a worthy enhancement.
> If I/someone does it, will it be merged in?

I really want it to happen, but I also want it to happen right.

See the relevant pull request[1].

[1] https://github.com/D-Programming-Language/druntime/pull/574
February 14, 2014
On Friday, 14 February 2014 at 09:30:30 UTC, bearophile wrote:
> Timothee Cour:
>
>> auto byKeyValue(){...}
>
> It's probably better to call the method "byPair" that is shorter.
>
>
>> If I/someone does it, will it be merged in?
>
> Some people want those pairs to be tuples. But D lacks built-in tuples, so you need to use typecons ones. But to use the typecons ones you need to import half Phobos. It seems there is no good solution.
>
> Bye,
> bearophile

I thought that the type of variadic template arg packs was considered D's tuple? The only problem is returning them from functions, but if that's fixed, would they not effectively be built-in tuples?
February 14, 2014
On Friday, 14 February 2014 at 15:36:30 UTC, Meta:

The only problem is returning them from
> functions, but if that's fixed, would they not effectively be built-in tuples?

It will not be fixed. Walter is against "fixing" that for efficiency reasons.

Bye,
bearophile