> may be something similar to Smaltalk block syntax: return(<var-name>,
<expression> [, <optional-message>])
That's a good approximation of perfect, I think ... looks okay to me.
Cheers
Joe
On Tue, Feb 10, 2015 at 10:21 AM, Dmitry Stogov <dmi...@zend.com> wrote:
> D uses block syntax:
>
> ensure(ret) {
> }
>
> In this case, passing 'ret' as a 'parameter' looks more or less natural.
>
> In our case, two sets of brackets looks weird.
> Predefined variable or constant looks better.
> A agree, __RETURN__ is longer than $ret, but it also won't conflict with
> other variables.
> I don't see a perfect solution yet.
>
> may be something similar to Smaltalk block syntax: return(<var-name>,
> <expression> [, <optional-message>])
>
> function add($a, $b)
> return($ret, $ret > 0)
> {
> return $a + $b;
> }
>
> Thanks. Dmitry.
>
> On Tue, Feb 10, 2015 at 1:03 PM, Joe Watkins <pthre...@pthreads.org>
> wrote:
>
>> The solution D uses is allowing the programmer to declare the name of the
>> return value like, we could do something like:
>>
>> public function method()
>> return ($result) ($result <= 10 && $result >= 0) {
>> return 5;
>> }
>>
>> If you didn't need the result, were working with a scope variable, or
>> maybe even a param taken by reference:
>>
>> public function method()
>> return ($expr) {
>> /* ... */
>> }
>>
>> I'm not keen on __CONSTANTS__, it's long, and if you have the choice to
>> name the variable you have the option of making the expression ...
>> expressive ...
>>
>> Cheers
>> Joe
>>
>> On Tue, Feb 10, 2015 at 9:48 AM, Dmitry Stogov <dmi...@zend.com> wrote:
>>
>>> I think __RETURN__ is better than $ret.
>>>
>>> Dmitry.
>>>
>>> On Tue, Feb 10, 2015 at 12:43 PM, Patrick Schaaf <p...@bof.de> wrote:
>>>
>>>> Am 10.02.2015 09:29 schrieb "Dmitry Stogov" <dmi...@zend.com>:
>>>> >
>>>> > I cnahged $> into $ret, because $> just won't work. Some better
>>>> solution is welcome.
>>>>
>>>> Also think $> would be awful. Don't like $ret either. What about a
>>>> magic constant __RETURN__ or __RESULT__? This would stand out very well
>>>> when glancing at the expression, would fit in with naming of other magic
>>>> constants, and might even be made to work in finally blocks, where the same
>>>> need for access to the returned value exists.
>>>>
>>>> best regards
>>>> Patrick
>>>>
>>>
>>>
>>
>
