Issue with Input file size and chunk size

[pbjarterot]

Hi, I've been trying to use build/compress with an example file i generated in python, as below. I keep getting the error

"Input file size is not a multiple of the chuink size"

i've tried many different values for -n but none seem to work, the last being

build/compress -i example_file.txt -o compressed_example_file.txt -n 512 1

Do you have a solution for this? Also, is it possible to provide a super simple example file to show how the input should look and how the command for said file should look so someone that is new to C++ tools like myself can get some more help

 import numpy as np

a = np.arange(1, 1024000, 1, dtype=float)

with open("example_file.txt", "w") as f:
    for i in a:
        f.write(f"{i}, ")

f.close() 

[fknorr]

Hi @pbjarterot, the input is expected to be a file of binary floating-point values, with 4 bytes per float for the default single-precision setting and 8 bytes per float for -t double.

From how I understand the numpy documentation, you should be able to output the desired format through ndarray.tofile().

-n takes one, two or three arguments for a one, two or three-dimensional input array respectively. For the file resulting from your array a this would be -n 1024000.

Hope that helps.

[pbjarterot]

Thank you, now it works, however, changing different values for n, i still get the ratio 1.0 every time, is this due to the type of data or any input argument?