heap allocate multi-dimensional array

[keen mulch]

How can i heap allocate for a 2 dimensional array? Im kinda confused.

perhaps the rows dimension becomes an arrays of pointers to the corresponding columns array?

[deep heart]

is it correct that you know the row-count and column-count when you want to allocate the array, but not at compile time?

[keen mulch]

yes, i know it at runtime, but not at compile time

Its a 2D array to represent the current screen of the terminal. I did this but im not sure if it will work. ```fn makeScreen(allocator: std.mem.Allocator, terminfo: termInfo) ![][]u8
{
var screen = try allocator.alloc(*[]u8, terminfo.rows);

var i = 0;
while (i != (terminfo.rows - 1)) : (i += 1)
{
    screen[i] = try allocator.alloc(u8, terminfo.columns);
}
return screen;

}```

[drowsy bison]

var arr: [][]u8 = undefined;
arr = try std.heap.page_allocator.alloc([]u8, 5);
for (0..5) |i| {
     arr[i] = try std.heap.page_allocator.alloc(u8, 4);
}

works but I think you can also use ptrcast an allocate once

[keen mulch]

this looks the most reasonable. Ill give it a try

[keen mulch]

oh and also, how can i infer the length of each dimension?

[deep heart]

I would allocate an array of size terminfo.rows * terminfo.columns. To a single row, you can slice the big array. This has the advantage, that you only have to allocate the array in one allocation function which is faster

[keen mulch]

...uhh what? How would i infer the elements then?

[drowsy bison]

arr.len and arr[0].len

[keen mulch]

thx, ill try to build my program on it and tell ya if it worked

[drowsy bison]

good luck. but a @deep heart suggested it is better to allocate once in terms of performance and complexity. You can define a custom structure to not write (i*cols + j) multiple times or you can allocate it one buffer just as suggested and create arr to point to the right elements. Here is a snippet if you want

var arr: [][]u8 = undefined;
const rows: u8 = 5;
const cols: u8 = 4;
const arr_tmp: []u8 = try std.heap.page_allocator.alloc(u8, rows * cols);
arr = try std.heap.page_allocator.alloc([]u8, rows);
for(0..rows) |i| {
  arr[i] = arr_tmp[i*cols .. (i+1)*cols];
}

but of course this uses more memory and needs to dereference twice to access data and you need to free arr_tmp and arr. It might not be a problem for you (I guess) but it is there to consider

and sorry for the bad format