`foreach_reverse` is allowed for delegates and do what `foreach` do thus causing dangerous confusion. Walter Bright is the only one for some obscure reason definitely for not deprecating this feature and WONTFIX-ed the issue. And yes, no big red note in docs about it. As always, this is even more error-prone for highly templated language like D (i.e. one can expect in template `foreach_reverse` will not compile if reverse iteration isn't supported).

Petition for all not indifferent to language misfeatures to chime in and show in this thread we all are for deprecating it (if we really are).

Issue URL: http://d.puremagic.com/issues/show_bug.cgi?id=1553

Thanks.

-- 
Денис В. Шеломовский
Denis V. Shelomovskij

20.10.2013 15:29, Denis Shelomovskij пишет:
> `foreach_reverse` is allowed for delegates and do what `foreach` do thus
> causing dangerous confusion. Walter Bright is the only one for some
> obscure reason definitely for not deprecating this feature and
> WONTFIX-ed the issue. And yes, no big red note in docs about it. As
> always, this is even more error-prone for highly templated language like
> D (i.e. one can expect in template `foreach_reverse` will not compile if
> reverse iteration isn't supported).
>
> Petition for all not indifferent to language misfeatures to chime in and
> show in this thread we all are for deprecating it (if we really are).
>
> Issue URL: http://d.puremagic.com/issues/show_bug.cgi?id=1553
>
> Thanks.

Also the funny thing is according to Andrei [1] about `foreach_reverse` over ranges from February 2010:
"Gone, that is. Walter agreed to remove it."

[1] http://forum.dlang.org/post/hl8345$2b1q$1@digitalmars.com

-- 
Денис В. Шеломовский
Denis V. Shelomovskij

Denis Shelomovskij:

> Issue URL: http://d.puremagic.com/issues/show_bug.cgi?id=1553

I suggest to keep foreach_reverse in the language, but statically disallow it on delegates.

Bye,
bearophile

20-Oct-2013 15:29, Denis Shelomovskij пишет:
> `foreach_reverse` is allowed for delegates and do what `foreach` do thus
> causing dangerous confusion. Walter Bright is the only one for some
> obscure reason definitely for not deprecating this feature and
> WONTFIX-ed the issue. And yes, no big red note in docs about it. As
> always, this is even more error-prone for highly templated language like
> D (i.e. one can expect in template `foreach_reverse` will not compile if
> reverse iteration isn't supported).
>
> Petition for all not indifferent to language misfeatures to chime in and
> show in this thread we all are for deprecating it (if we really are).
>

+1


At a times it looks as if Walter is in favor of keeping whatever broken junk forever just because it was always broken.

> Issue URL: http://d.puremagic.com/issues/show_bug.cgi?id=1553
>
> Thanks.
>


-- 
Dmitry Olshansky

On 10/20/2013 01:29 PM, Denis Shelomovskij wrote:
> `foreach_reverse` is allowed for delegates and do what `foreach` do thus
> causing dangerous confusion. Walter Bright is the only one for some
> obscure reason definitely for not deprecating this feature and
> WONTFIX-ed the issue.  ...

This is by design? I thought it was obviously a bug in the implementation.

On Sunday, 20 October 2013 at 12:33:27 UTC, bearophile wrote:
> Denis Shelomovskij:
>
>> Issue URL: http://d.puremagic.com/issues/show_bug.cgi?id=1553
>
> I suggest to keep foreach_reverse in the language, but statically disallow it on delegates.

+1

foreach_reverse is ugly, but works, and there's no point breaking code just to tidy things up a little.

foreach_reverse on delegates on the other hand is completely broken, so should be statically disallowed (preferably with a nice error message explaining that the current behaviour is broken!)

On Sunday, 20 October 2013 at 11:29:38 UTC, Denis Shelomovskij wrote:
> `foreach_reverse` is allowed for delegates and do what `foreach` do thus causing dangerous confusion. Walter Bright is the only one for some obscure reason definitely for not deprecating this feature and WONTFIX-ed the issue. And yes, no big red note in docs about it. As always, this is even more error-prone for highly templated language like D (i.e. one can expect in template `foreach_reverse` will not compile if reverse iteration isn't supported).
>
> Petition for all not indifferent to language misfeatures to chime in and show in this thread we all are for deprecating it (if we really are).
>
> Issue URL: http://d.puremagic.com/issues/show_bug.cgi?id=1553
>
> Thanks.

I don't use foreach_reverse and don't intend to. I allowing delegates is error prone if it is just going to do what foreach does.

Am Sun, 20 Oct 2013 15:55:41 +0200
schrieb "Peter Alexander" <peter.alexander.au@gmail.com>:

> On Sunday, 20 October 2013 at 12:33:27 UTC, bearophile wrote:
> > Denis Shelomovskij:
> >
> >> Issue URL: http://d.puremagic.com/issues/show_bug.cgi?id=1553
> >
> > I suggest to keep foreach_reverse in the language, but statically disallow it on delegates.
> 
> +1

+1

Often enough I see code that iterates forwards and backwards
over something. foreach and foreach_reverse make the syntax
look the same. I don't like the idea of plugging in
reverse ranges into foreach to get the same effect.

-- 
Marco