You are not logged in.
The script never gives me servers in the best order for speed, regardless of how many geographic regions I omit from the pacman repo mirror lists. In fact, it regularly reckons that the 'best' servers are some of the slowest in terms of download speeds.
I'm all in favour of having automated server benchmarking, it's just the script perhaps biases speed of establishing connection over speed of the connection itself...
- Dave
Last edited by KomodoDave (2007-06-23 11:12:12)
Offline
Feel free to look at how it works, and see if you find a way to improve it.
It's just 150 lines of python. I don't even know python, but it doesn't look very complicated
It apparently just checks the time for downloading the first 50kb :
# retrieve first 50,000 bytes only
tmp = opener.open(serverUrl).read(50000)
Maybe for a small amount of data like this, the speed of establishing connection is more important than the speed of the connection itself.
It's maybe possible to establish the connection first, then time only the download, I don't know.
Or just adding an argument to the script for using a bigger value of data to download, to see if it helps.
Anyway, it works good enough for me, since the first two mirrors it reports are close, respond fast, and are fast too (or at least, enough for my connection which is at ~ 1.2 MB/s ).
pacman roulette : pacman -S $(pacman -Slq | LANG=C sort -R | head -n $((RANDOM % 10)))
Offline