"Denis Koroskin" <2korden@gmail.com> wrote in message news:op.ur2f3rglo7cclz@korden-pc...
> On Wed, 08 Apr 2009 15:04:29 +0400, Frits van Bommel <fvbommel@remwovexcapss.nl> wrote:
>
>> Andrei Alexandrescu wrote:
>>> Denis Koroskin wrote:
>>>> Well, actually I think that having an Appender object is an overkill. I never use, although I wrote a few implementations. Instead, I found the following method to be extemely handy, very fast and cover all my cases:
>>>>
>>>> void append(T)(T[] array, ref size_t index, T value)
>>>> {
>>>>    assert(array.length >= index);
>>>>    if (array.length == index) {
>>>>        array.length = array.length * 2;
>>>>    }
>>>>        array[index++] = value;
>>>> }
>>>  I'm pretty sure you meant to pass array by reference.
>>
>> It also breaks when the array is empty.
>
> Yeah, I was writing from memory and could (and did!) introduce bugs.
>
> My intend was to show an easy way of appending to array without use of a special Appender struct. I use it /very/ often and believe it belongs to std.array.
>

A lot of code doesn't expect extra "unofficial" elements at the end of an array. Any such code would break if that append had been used on the array it's operating on. In my experience, such code occurs often enough to make that a dangerous strategy. You could be careful and make sure all your code is designed in a way that accomodates such extra "allocated-but-unused" elements, but for all the bother that would be, you might as well just use a real .length/.capacity solution.

On Fri, 10 Apr 2009 11:00:01 +0400, Nick Sabalausky <a@a.a> wrote:

> "Denis Koroskin" <2korden@gmail.com> wrote in message
> news:op.ur2f3rglo7cclz@korden-pc...
>> On Wed, 08 Apr 2009 15:04:29 +0400, Frits van Bommel
>> <fvbommel@remwovexcapss.nl> wrote:
>>
>>> Andrei Alexandrescu wrote:
>>>> Denis Koroskin wrote:
>>>>> Well, actually I think that having an Appender object is an overkill.
>>>>> I never use, although I wrote a few implementations. Instead, I found
>>>>> the following method to be extemely handy, very fast and cover all my
>>>>> cases:
>>>>>
>>>>> void append(T)(T[] array, ref size_t index, T value)
>>>>> {
>>>>>    assert(array.length >= index);
>>>>>    if (array.length == index) {
>>>>>        array.length = array.length * 2;
>>>>>    }
>>>>>        array[index++] = value;
>>>>> }
>>>>  I'm pretty sure you meant to pass array by reference.
>>>
>>> It also breaks when the array is empty.
>>
>> Yeah, I was writing from memory and could (and did!) introduce bugs.
>>
>> My intend was to show an easy way of appending to array without use of a
>> special Appender struct. I use it /very/ often and believe it belongs to
>> std.array.
>>
>
> A lot of code doesn't expect extra "unofficial" elements at the end of an
> array. Any such code would break if that append had been used on the array
> it's operating on. In my experience, such code occurs often enough to make
> that a dangerous strategy. You could be careful and make sure all your code
> is designed in a way that accomodates such extra "allocated-but-unused"
> elements, but for all the bother that would be, you might as well just use a
> real .length/.capacity solution.
>
>

Real .length and .capacity solution would be the best solution, but since slices currently lack one of these properties, I carry length outside and re-use buffer.length as capacity.

In most cases, I use this pattern for appending to a temporary buffer (which is designed to be re-used, even if it stores something). And if it stores values, then it's either 1) full and you just set "index = buffer.length;" at the beginning, or 2) size is provided ("index = initialSize;").

Never experienced any problems with it.