The app prints every size in GB, so most findings read 0.00 GB

[Muawiya-contact]

The desktop app has two size formatters and both are fixed to one unit:

function bytesToGB(n) {
  return (n / 1e9).toFixed(2);
}

...used for every headline and group total, and (bytes / 1e6).toFixed(1)

• " MB" for every row. So the numbers the app shows are:

bytes	headline	row
2 048	0.00 GB	0.0 MB
1 500 000	0.00 GB	1.5 MB
40 000 000	0.04 GB	40.0 MB
2 300 000 000	2.30 GB	2300.0 MB

A scan of a home directory that frees 40 MB reports 0.04 GB reclaimable,
every cache file under a megabyte reads 0.0 MB, and a 2.3 GB installer
is a four-digit MB number. "Understand your disk before you clean it"
should not need the reader to count decimal places.

The CLI already solved this — human_bytes in
crates/diskern-cli/src/main.rs, including the rounding detail that
prints 999 999 as 1.0 MB rather than 1000.0 KB. The app wants the same
function in JS, used everywhere a byte count is printed, so the two
frontends describe the same finding the same way.

Call sites: app/src/App.jsx:26 (bytesToGB), and the raw / 1e6 and
/ 1e9 at lines 94, 144, 181, 189, 190, 261, 361, 546.

[haas26p-ctrl]

I'd like to take this.

Plan: replace bytesToGB and the fixed / 1e6 / / 1e9 call sites in app/src/App.jsx with a JS humanBytes that mirrors human_bytes in crates/diskern-cli (SI units, 999.95 carry so 999_999 → 1.0 MB).