On Wed, 29 Apr 2015, jb wrote: > I agree! but I don't want to rush. What latency should substitute? average? > median? latency during upload? download? both? should it be the excess over > idle ping, expressed as a multiplier? an absolute value? an absolute value of excess over idle (see the other discussion about jitter) median +- range would be ideal as that could cover both upload and download > Most people as in 98% of people who use this test do so without any > reference to instructions the forums or anything else and all of them > "expect" to see latency > pings that are similar to speedtest.net measure of how close the speed test > server is. In this case, we need something like: latency base + value[+-] loaded David Lang > And I haven't even sorted out the latency bugs. I just now asked the nginx > development list why IE11 has problems doing http pings because it does > a weird two-step of high-normal-high-normal, but when talking to nginx only! > it behaves ok when talking to google.com. > > There is a jsfiddle you can play with to see the problem (I hope!) > http://jsfiddle.net/qe44nbwh/ > change nginx.org to google.com, using IE11. (other browsers work fine) > > On Wed, Apr 29, 2015 at 12:23 AM, Dave Taht wrote: > >> http://dfkrkqaqb1zsx.cloudfront.net/speedtest/cdn/377727.png >> >> is showing the base latency without load. It would be more effective >> if it showed the >> latency with load. >> >> -- >> Dave Täht >> Open Networking needs **Open Source Hardware** >> >> https://plus.google.com/u/0/+EricRaymond/posts/JqxCe2pFr67 >> >