This looks good to me and is probably more true to how most bandwidth is recorded, but we should probably change to the proper units as well (eg "MB" -> "MiB").

As I understand, things like MB and KB are the SI units divided just by things like 1_000_000 where the binary versions like 1_048_576 are Mebibytes as opposed to Megabytes officially. They end up being mixed up a lot I think, but it wouldn't hurt to add the extra "i" to be clear. So things like "KiB", "MiB", "GiB", and "TiB".

Otherwise, looks like a good change! It could be extra work, but we could also have a flag for picking the units? The only reason I say that is it may be confusing if you download a 1GB file and see something like 0.93GiB in the cumulative mode

That's interesting. Hum... FF and wget seem to be reporting these as {M,K}B/s. Running some manual tests, it seems to me* that we match both of these. So do you think this is a "mistake" on their part, or just an omission to make things simple?

*hard to be 100% sure because we're averaging our reporting over 3 seconds

Did some tests with -t and we're really dead-on with what wget reports. So I guess they're reporting mebibytes too. I'll change everything to {mebi}bytes. Learned something new - thanks! :)

I dunno about the flag. I want to try to avoid adding more flags tbh. Maybe if someone specifically needs the distinction?

Haha, upon looking into this further, it seems like a bit of a mess. From what I can gather, FF uses the powers-of-two even when the operating system (like Linux and MacOS, I believe) use the SI powers-of-ten convention. Here is a bug report about that and an ancient, still open one on a similar topic.

I think wget does the same?

I think though, that this is the standard the Unixes have settled on for the most part?

And yeah, on second thought, it's probably not worth the flag. I think that moving to the "ibi" units is nice just to be totally unambiguous :)

Alright, changed everything to ibis. :) Going to merge this. Thanks for the valuable input!