per_sec and eta values fluctuate wildly from one tick to the next

[shssoichiro]

It seems that this originated in 71e7eaa.

The previous behavior was that the per_sec and eta values would be relatively steady throughout the life of the progress bar, as they simply calculated pos / (now - start_time).

The new behavior instead calculates (pos - prev_pos) / (now - prev_time). Since this no longer has the effect of averaging out fast and slow spikes since the start of the progress bar, it causes the values in per_sec and eta to fluctuate wildly between each tick. In my opinion, this is undesirable.

[djc]

I think this is the discussion in #394. It could really use someone to dig in and come up with a good plan, and I'm not sure I'll have time for that in the near term.

[teor2345]

Closed by #539?

[djc]

Probably! If not, feel free to reopen.