At 11:45 +0200 6/8/09, Daniel Stenberg wrote, and I snipped:
>The average speed that curl shows is the average speed so far during the entire transfer. Going down 2 out of 1800 is about a tenth of a percent. It's not like it is a significant amount that I find it worth to lose a lot of sleep over.

As a certified quantum mechanic I think the problem may well be an example of the uncertainty principle. Those one-per-second calculations are triggered by some kind of interrupt that is not phase related to the passage of information in another thread. It may well be impossible to get a precise status of the data stream when activation of the progress code is a bit like opening Schroedinger's box of bits. While the calculation is proceeding the it count may, and may not, get incremented.

The overall average rate ought to be the current total bit count divided by the total time. Is it? I spend a significant fraction of my life teaching that one should never average a list of averages,

The world needs more folks like Eric who notice little things like that.

-- 
--> If you are presented a number as a percentage, and you do not clearly understand the numerator and the denominator involved, you are surely being lied to. <--
-------------------------------------------------------------------
List admin: http://cool.haxx.se/cgi-bin/mailman/listinfo/curl-users
FAQ:        http://curl.haxx.se/docs/faq.html
Etiquette:  http://curl.haxx.se/mail/etiquette.html