## digitalmars.D.learn - Synax for variadic template

• Alex (30/30) Sep 01 2017 Hi all!
• Nicholas Wilson (10/40) Sep 01 2017 b.arr refers to an `(AliasSeq!(int, double))[]`, so with
• Alex (9/18) Sep 01 2017 Ok... I think I need some type of soa, yes.
```Hi all!
Say, I have

struct A(T...)
{
T arr;
}

struct B(T...)
{
T[] arr;
}

void main()
{
A!(int[], double[]) a;
a.arr ~= 5;
a.arr ~= 6;
static assert(!__traits(compiles, a.arr ~= 3.5));
a.arr ~= 19.8;

assert(a.arr == [5,6]);
assert(a.arr == [19.8]);
assert(a.arr.length == 1);
assert(a.arr.length == 2);

B!(int, double) b;
//b.arr ~= 5;
}

While struct A behaves like expected, how to get by with B?
What I want is just to abstract the array property out of
template arguments...

```
Sep 01 2017
```On Friday, 1 September 2017 at 09:38:59 UTC, Alex wrote:
Hi all!
Say, I have

struct A(T...)
{
T arr;
}

struct B(T...)
{
T[] arr;
}

void main()
{
A!(int[], double[]) a;
a.arr ~= 5;
a.arr ~= 6;
static assert(!__traits(compiles, a.arr ~= 3.5));
a.arr ~= 19.8;

assert(a.arr == [5,6]);
assert(a.arr == [19.8]);
assert(a.arr.length == 1);
assert(a.arr.length == 2);

B!(int, double) b;
//b.arr ~= 5;
}

While struct A behaves like expected, how to get by with B?
What I want is just to abstract the array property out of
template arguments...

b.arr refers to an `(AliasSeq!(int, double))[]`, so with
`b.arr ~= 5;` you are trying to append a integer to an array
of pairs of ints and doubles, which you can't do.

b.arr ~= ElementType!(typeof(b.arr))(5,42.0);

should work (I hope, not tested) but you cannot append only the
int.

If you are trying to abstract the array for struct of array vs.
array of struct the take a look at
https://maikklein.github.io/post/soa-d/
```
Sep 01 2017
```On Friday, 1 September 2017 at 10:01:16 UTC, Nicholas Wilson
wrote:
b.arr refers to an `(AliasSeq!(int, double))[]`, so with
`b.arr ~= 5;` you are trying to append a integer to an array
of pairs of ints and doubles, which you can't do.

b.arr ~= ElementType!(typeof(b.arr))(5,42.0);

should work (I hope, not tested) but you cannot append only the
int.

If you are trying to abstract the array for struct of array vs.
array of struct the take a look at
https://maikklein.github.io/post/soa-d/

Ok... I think I need some type of soa, yes.
But in the article, there is too much of effort done, to convert
types to arrays, at least at the stage of my code... At the end,
if I compare the amount of code in the article and just "[]"
more, as for the part of my example that works...

Thanks for the cool link, nevertheless... Cool to see, how it
should be done :)
```
Sep 01 2017