Generic range to read array or a file as a phobos range

[bioinfornatics]

On Sunday, 17 March 2013 at 20:21:23 UTC, Andrea Fontana wrote:
> On Sunday, 17 March 2013 at 18:36:17 UTC, bioinfornatics wrote:
>> On Friday, 15 March 2013 at 16:40:26 UTC, Andrea Fontana wrote:
>>> On Friday, 15 March 2013 at 16:11:38 UTC, bioinfornatics 
>>> wrote:
>>>> On Friday, 15 March 2013 at 14:40:17 UTC, Andrea Fontana 
>>>> wrote:
>>>>> On Friday, 15 March 2013 at 14:31:43 UTC, bioinfornatics 
>>>>> wrote:
>>>>>> Dear,
>>>>>>
>>>>>> By using CTFE I try to get a generic range to read array 
>>>>>> or a file as a phobos range. code hosted here: 
>>>>>> http://dpaste.dzfl.pl/1f2bcf39
>>>>>>
>>>>>> that works fine for array but for a File instance .eof 
>>>>>> seem to not return true a right time.
>>>>>>
>>>>>> Soemone could say what happen ?
>>>>>>
>>>>>> Thanks
>>>>>
>>>>> Have you tried to cache front() result and read next block 
>>>>> on popFront()?
>>>>
>>>> that is ok. Just missed the \n the followed code should work 
>>>> http://dpaste.dzfl.pl/1f2bcf39
>>>>
>>>> You are welcome to take it :)
>>>
>>> int[] a = [ 0, 1, 1, 2, 3, 5, 8 ];
>>> std.file.write("/tmp/filename", a);
>>>
>>> File f = File("/tmp/filename", "r");
>>> auto r = GenericRange!File(f,1);
>>>
>>> r.filter!"a.length > 0 && a[0] != 0"().writeln;
>>>
>>> output:
>>> [[0], [0], [0], [0], [0], [0]]
>>>
>>> but i think output expected is:
>>> [[1], [1], [2], [3], [5], [8]]
>>>
>>> Caching front() and reading buffer from popFront() did the 
>>> trick for me.
>>
>> As you said that do not works i update the code with some
>> unittest could you please to share your version ? you can use
>> fork from dpast -> http://dpaste.dzfl.pl/1f2bcf39e. Thanks
>
> Your code works if you use .writeln instead of .array. This why 
> *it seems* that struct member "U buffer" is returned by 
> reference.
>
> Check this. It passes unittest.
> http://dpaste.dzfl.pl/828f7cc4
>
> I move buffer inside popFront function, so "value = 
> rawWrite(..)" assign every time a reference to a different 
> "buffer". In your code it returns always the same member 
> variable so, output array was an array of 8 identical reference 
> to last read value (e.g. [8]) Maybe you can achieve the same in 
> other way (for example using idup?).
>
> I hope it works.

yes that was a reference problem using dup fix it. Thanks