Description

I can't believe that hyfetch has been showing wrong (inaccurate at best) values for used/total memory for this long. And the memory precision feature seems broken from the beginning.

The problem: if we had 15,361 to 16,383 KiB of memory, hyfetch would show 15.00 MiB instead of the correct decimal value, because it represented memory values in integral MiB internally. If we had 8,192 MiB of memory, hyfetch would show 0.819 TiB (!) instead of 0.008 TiB (if you chose that unit), because the algorithm for digits after the decimal point was incorrect.

To fix the problem, we change to KiB internal representation (basically remove one division by 1024 from raw values), and we rewrite the assignment for final mem_used and mem_total strings. Still no awk involved!

Features

Issues

• If bar_length were very large (>4000), it might overflow on 32-bit machines.
• If mem_precision were large (>=10 on 32-bit or >=19 on 64-bit), it would overflow.

TODO

1. Banker's rounding instead of rounding down
2. Better handling for insane config values/command arguments (not a problem specific for this feature, though)