Open Bug 1281668 Opened 8 years ago Updated 2 years ago

The download speed/time computation for each individual download is incorrect

Categories

(Firefox :: Downloads Panel, defect, P5)

39 Branch
defect

Tracking

()

People

(Reporter: gvenkat, Unassigned)

Details

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.11; rv:39.0) Gecko/20100101 Firefox/39.0
Build ID: 20150806001005

Steps to reproduce:

Download a large file from a slow server so it takes at least 10s of minutes to observe.

Case 1: Just this file downloading in the current session.

Case 2: Download another file from a fast server while the above is in progress OR before it so it completes quickly with a much higher speed.


Actual results:

Case 1: The time and speed is reported reasonably well even with variations in speed

Case 2: The time and speed in real time is completely wrong for the slow download in the download list. What seems to have happened is that the high speed of the faster download is taken into the calculation of the slower download which makes no sense. So the speed reported for the slower download will become much higher than what it is for that download even if it is the only one at the moment (after the faster download has completed) and so the time computed is nonsensical, has no relation to how much time it will take (even IF the download speed is extremely steady)


Expected results:

The speeds for each individual download should be computed for bytes received for that download only not mixed with other downloads in the same session, especially since these times are reported individually for each separate download in progress. The current estimates make absolutely no sense.
Component: Untriaged → Downloads Panel
Thanks for reporting this issue. We use a simple algorithm to calculate the download speed and it may definitely have some limitations like the one you described. Each speed is however calculated individually, so I'm not sure why the presence of other downloads affects it, it might be a general issue with the two downloads sharing the same bandwidth.

I don't have the theoretical knowledge to improve the situation, but if you have any example of other programs or algorithms doing a better job in the case you described, we might definitely investigate and adopt them.
Status: UNCONFIRMED → NEW
Ever confirmed: true
Priority: -- → P5
This could be a duplicate of bug 435380: the fast download could be saturating your connection and making the slow download more bursty. The current algorithm is very inaccurate for bursty downloads.
Severity: normal → S3
You need to log in before you can comment on or make changes to this bug.