Describe the bug

I'm currently writing a command that performs an action every 2048 bytes of binary data. That's when I noticed that the command takes longer, the bigger the input data is. This sounds like the binary data is first loaded entirely and then sliced. Taking the first 2048 bytes of data should always take the same amount of time, regardless of input size.

How to reproduce

1. Create a file that is 500 KB
2. Create a file that is 500 MB
3. let small_input = (open --raw 500kb.bin | into binary)
4. let big_input = (open --raw 500mb.bin | into binary)
5. timeit { $small_input | bytes at ..2048 }
6. timeit { $big_input | bytes at ..2048 } --> will take around 1000 times longer

Expected behavior

Should always take the same amount of time. Comparable tools work like this:

• timeit { head -c 2048 500kb.bin } and timeit { head -c 2048 500mb.bin }
• timeit { od -j 20000 -N 2048 500kb.bin } and timeit { od -j 20000 -N 2048 500mb.bin } (get 2048 bytes starting from byte 20000)

These commands don't take longer to execute if the input is bigger in size. Even slices from the end of the file don't take longer.

Screenshots

No response

Configuration

Additional context

No response