On Sunday, October 11, 2015 05:10:34 tcak via Digitalmars-d-learn wrote:
> On Saturday, 10 October 2015 at 20:07:11 UTC, Jonathan M Davis
> wrote:
> > On Saturday, October 10, 2015 15:20:02 tcak via
> > Digitalmars-d-learn wrote:
> >> [code]
> >> int[] list;
> >>
> >> list = new int[0];
> >>
> >> std.stdio.writeln("Is Null ? ", (list is null));
> >> [/code]
> >>
> >> Result is "Is Null? true".
> >>
> >> Is this the correct behaviour? I would expect compiler to
> >> point to an address in the heap, but set the length as 0. So,
> >> it wouldn't return null, but the length would be 0 only.
> >
> > It basically didn't bother to allocate an array on the heap,
> > because you asked for one with a length of zero.
> > Efficiency-wise, it makes no sense to allocate anything. You
> > wouldn't be doing anything with the memory anyway. The only way
> > that you're going to get an array of length 0 which doesn't
> > have a null ptr is to slice an array down to a length of 0.
> >
> > - Jonathan M Davis
>
> The situation is that the "length" parameter comes from user.
> Also the item values come from user as well. I create the array
> with "length" parameter. At another part of code, I check firstly
> whether the array is created [code] if( array is null ) [/code],
> then the items are checked for validation.
In general, because of how arrays tend to conflate null and empty, it's a
bad idea to differentiate between null and empty with arrays. I don't know
exactly what you're doing, but there's no reason to check for null before
iterating over an array, because null == "" and null == []. You'll never get
a segfault from operating on a null array unless you try and do something
with its ptr property explicitly. Almost everything treats a null array the
same as an empty array. If you really need to have a null value for arrays,
consider using std.typecons.Nullable to wrap the array.
- Jonathan M Davis
