On 2010-12-29 at 11:31:33 [+0100], Oliver Tappe <zooey@xxxxxxxxxxxxxxx> wrote: > [ ... ] > > Anyway, I'm going to test the performance difference between GA active and > inactive today, so I can give you some numbers. I have done that now and the results are that GA doesn't influence the performance in any meaningful manner, at least I wasn't able to read any relevant difference from the two jmeter runs. Dunno why I got the impression that it slowed things down in the first place, I must have mixed up something when experimenting with deactivating all those Drupal modules. So, from a performance viewpoint, GA can be left active on our real site. Additionally, I have looked at our GA stats and they don't seem to be much different from what other log analyzers produce, albeit a bit more detailed (and reliable, I suppose, as they have code implanted in the browsers). The concrete numbers are greatly different from the ones shown on our web-stats page, but I haven't any idea about why that is, yet. I'm no expert on this stuff, but looking at the data GA exchanges with its server, I get the impression that they are able to track user specific actions across sites with GA. As that's not quite my cup of tea, I tried to get rid of the utm... cookies, but with Firefox I wasn't able to do that unless I chose to reject all cookies of www.haiku-os.org. That of course would include the session cookie, so that doesn't make much sense. Bummer ... cheers, Oliver ----------------------------------------------------------------------- haiku-web@xxxxxxxxxxxxx - Haiku Web & Developer Support Discussion List