Consider:

auto s = "some string";
static assert(!hasLength(typeof(s));
auto m = map!(a => a)(s);
static assert(!hasLength(typeof(s));

Both asserts should pass: string famously has a length that's actually not the length of the range. Then mapping something on top of the string should also have no meaningful length. Alas, it does, because of the following code in std.algorithm:

static if (hasLength!R || isSomeString!R)
{
    @property auto length()
    {
         return _input.length;
    }
    alias length opDollar;
}

The code used to introduce length only if if (hasLength!R).

It seems Kenji introduced this change. What was its purpose, and can we undo it? This breaks the entire web of assumptions that hasLength builds.


Thanks,

Andrei
_______________________________________________
phobos mailing list
phobos@puremagic.com
http://lists.puremagic.com/mailman/listinfo/phobos

On 6 aug 2012, at 20:55, Andrei Alexandrescu wrote:

> Consider:
> 
> auto s = "some string";
> static assert(!hasLength(typeof(s));
> auto m = map!(a => a)(s);
> static assert(!hasLength(typeof(s));


Do you mean:

static assert(!hasLength(typeof(m)));

For the second assert?

-- 
/Jacob Carlborg

_______________________________________________
phobos mailing list
phobos@puremagic.com
http://lists.puremagic.com/mailman/listinfo/phobos

On 8/6/12 3:11 PM, Jacob Carlborg wrote:
>
> On 6 aug 2012, at 20:55, Andrei Alexandrescu wrote:
>
>> Consider:
>>
>> auto s = "some string";
>> static assert(!hasLength(typeof(s));
>> auto m = map!(a =>  a)(s);
>> static assert(!hasLength(typeof(s));
>
>
> Do you mean:
>
> static assert(!hasLength(typeof(m)));
>
> For the second assert?

Yes, apologies. Essentially mapping a function over characters in a string should not automagically define a length property.

Andrei

_______________________________________________
phobos mailing list
phobos@puremagic.com
http://lists.puremagic.com/mailman/listinfo/phobos

On Monday, August 06, 2012 14:55:56 Andrei Alexandrescu wrote:
> It seems Kenji introduced this change. What was its purpose, and can we undo it? This breaks the entire web of assumptions that hasLength builds.

Then create a bug report and mark it as a regression. I don't think that there's any real question that it's wrong.

- Jonathan M Davis
_______________________________________________
phobos mailing list
phobos@puremagic.com
http://lists.puremagic.com/mailman/listinfo/phobos

It seems to me that is an old bug. https://github.com/D-Programming-Language/phobos/commit/3e6679b2973069ba9191da16fd74d8b8b80865eb#L1L437

Regards.

Kenji Hara

2012/8/7 Andrei Alexandrescu <andrei@erdani.com>:
> Consider:
>
> auto s = "some string";
> static assert(!hasLength(typeof(s));
> auto m = map!(a => a)(s);
> static assert(!hasLength(typeof(s));
>
> Both asserts should pass: string famously has a length that's actually not the length of the range. Then mapping something on top of the string should also have no meaningful length. Alas, it does, because of the following code in std.algorithm:
>
> static if (hasLength!R || isSomeString!R)
> {
>     @property auto length()
>     {
>          return _input.length;
>     }
>     alias length opDollar;
> }
>
> The code used to introduce length only if if (hasLength!R).
>
> It seems Kenji introduced this change. What was its purpose, and can we undo it? This breaks the entire web of assumptions that hasLength builds.
>
>
> Thanks,
>
> Andrei
> _______________________________________________
> phobos mailing list
> phobos@puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
_______________________________________________
phobos mailing list
phobos@puremagic.com
http://lists.puremagic.com/mailman/listinfo/phobos

On 06-Aug-12 22:55, Andrei Alexandrescu wrote:
> Consider:
>
> auto s = "some string";
> static assert(!hasLength(typeof(s));
> auto m = map!(a => a)(s);
> static assert(!hasLength(typeof(s));
>
> Both asserts should pass: string famously has a length that's actually not the length of the range. Then mapping something on top of the string should also have no meaningful length. Alas, it does, because of the following code in std.algorithm:
>
> static if (hasLength!R || isSomeString!R)
> {
>     @property auto length()
>     {
>          return _input.length;
>     }
>     alias length opDollar;
> }
>
> The code used to introduce length only if if (hasLength!R).
>
> It seems Kenji introduced this change. What was its purpose, and can we undo it? This breaks the entire web of 
It bugs me  - the indecision on this matter. No Kenji didn't introduce it.

The guilty commit (that could be reverted on it's own):

SHA-1: d0457fe82a5f140218d8a9700eb7c26a236d4abe

* This test is actually needed

Adds "isSomeString" to map's "length" test.
Author: monarchdodra

Contains one line diff that goes as follows:
-static if (hasLength!R)
+static if (hasLength!R || isSomeString!R)

P.S. I do not suggest to gill our new contributors but to pay more attention during the review of pulls.

-- 
Olshansky Dmitry

_______________________________________________
phobos mailing list
phobos@puremagic.com
http://lists.puremagic.com/mailman/listinfo/phobos