Re: Unmanaged drop in replacemet for [] and length -= 1
On Monday, 20 June 2016 at 16:27:29 UTC, Steven Schveighoffer
wrote:
On 6/18/16 5:55 PM, Joerg Joergonson wrote:
I wanted to switch to std.container.Array but it doesn't seem
to mimic
[] for some odd ball reason. I threw this class together and
it seems to
work.
The only problem is that I can't do
carray.length -= 1;
I can't override `-=` because that is on the class. can I
override it
for length somehow or do I have to create a length wrapper
class that
has it overridden in it? Or is there a way to do it in cArray?
length wrapper *struct*:
struct AdjustableLength
{
cArray t;
auto get() { return t.data.length; }
opOpAssign(string s: "+", Addend)(Addend x)
{
//... your code here that does += using t
t.data.length = get() + x;
}
alias get this;
}
@property auto length()
{
return AdjustableLength(this);
}
D does not have any direct support for modification of
properties. It has been talked about, but has never been
implemented.
-Steve
Thanks, this is what I was thinking I'd have to do. I'm probably
going to manually manage the array items. I just need simple
append and remove but this will still help with the drop in
replacement(for my cases).
Re: Unmanaged drop in replacemet for [] and length -= 1
On 6/18/16 5:55 PM, Joerg Joergonson wrote:
I wanted to switch to std.container.Array but it doesn't seem to mimic
[] for some odd ball reason. I threw this class together and it seems to
work.
The only problem is that I can't do
carray.length -= 1;
I can't override `-=` because that is on the class. can I override it
for length somehow or do I have to create a length wrapper class that
has it overridden in it? Or is there a way to do it in cArray?
length wrapper *struct*:
struct AdjustableLength
{
cArray t;
auto get() { return t.data.length; }
opOpAssign(string s: "+", Addend)(Addend x)
{
//... your code here that does += using t
t.data.length = get() + x;
}
alias get this;
}
@property auto length()
{
return AdjustableLength(this);
}
D does not have any direct support for modification of properties. It
has been talked about, but has never been implemented.
-Steve
Re: Unmanaged drop in replacemet for [] and length -= 1
On Sunday, June 19, 2016 15:59:41 Joerg Joergonson via Digitalmars-d-learn wrote: > If foreach removes all/any of the elements of a container then > something is broke. That's exactly what happens with a basic input range, and if it doesn't happen with a forward range, it's just because copying that range implicitly saves it, which doesn't work in generic code, because it's not the case with all forward ranges. popFront consumes an element from a range, and foreach is popping every element in the range. The same goes for any algorithm which iterates over a range. So, naturally, any container that is treated as a range is going to have its elements removed as it's iterated over. You're certainly free to treat a container as a range if that's what you want to do, but it shows a fundamental misunderstanding of what a range is supposed to be and do, and there are plenty of algorithms that will behave very badly for you if you do that. - Jonathan M Davis
Re: Unmanaged drop in replacemet for [] and length -= 1
On Sunday, 19 June 2016 at 10:10:54 UTC, Jonathan M Davis wrote: On Saturday, June 18, 2016 21:55:31 Joerg Joergonson via Digitalmars-d-learn wrote: I wanted to switch to std.container.Array but it doesn't seem to mimic [] for some odd ball reason. D's dynamic arrays are really quite weird in that they're sort of containers and sort of not. So, pretty much nothing is ever going to act quite like a dynamic array. But when dynamic arrays are used as ranges, their semantics definitely are not that of containers. Having a container which is treated as a range is just begging for trouble, and I would strongly advise against attempting it. When it comes to containers, ranges are intended to be a view into a container, just like an iterator is intended to be a pointer into a container. Neither ranges are iterators are intended to _be_ containers. Treating a container as a range is going to get you weird behavior like foreach removing every element from the container. - Jonathan M Davis Thanks for your 2c... but I think I can handle it. I'm a big boy and I wear big boy pants and I'm not afraid of a few little scrapes. If foreach removes all/any of the elements of a container then something is broke.
Re: Unmanaged drop in replacemet for [] and length -= 1
On Saturday, June 18, 2016 21:55:31 Joerg Joergonson via Digitalmars-d-learn wrote: > I wanted to switch to std.container.Array but it doesn't seem to > mimic [] for some odd ball reason. D's dynamic arrays are really quite weird in that they're sort of containers and sort of not. So, pretty much nothing is ever going to act quite like a dynamic array. But when dynamic arrays are used as ranges, their semantics definitely are not that of containers. Having a container which is treated as a range is just begging for trouble, and I would strongly advise against attempting it. When it comes to containers, ranges are intended to be a view into a container, just like an iterator is intended to be a pointer into a container. Neither ranges are iterators are intended to _be_ containers. Treating a container as a range is going to get you weird behavior like foreach removing every element from the container. - Jonathan M Davis
Re: Unmanaged drop in replacemet for [] and length -= 1
Also, how to handle foreach(i, x; w) (use index + value)?
