>>I feel the idea is not accurate. My ip address (a static one from my 
>>isp) would remain the same if I went to ADSL from my current 
>>dial-up access.
> Yes. And how about all corporate networks? If I'm accessing your site from
> our office I would get the low-speed version until you added our domainname
> to your list. Then what if I connect to our modem-pool from home... Then you
> wouldn't know I connected through a modem.
> And my ISP at home doesn't have anything in their dns-names to differentiate
> adsl from dial-up users.
> Just my thoughts...
> Regards
> Joakim Andersson

Yes, although the problem of determining users' bandwidth *is* real, I 
guess solution is not to be found in whatver IP address processing you 
might do.

But you might consider sampling your actual communication, like sending 
the browser a sample document of fixed length that will execute some 
answer action from the browser. If you can time the lenght of this loop 
you can have a rought idea of what the user bandwidth is at the moment.

Besides, no matter what your bandwidth to your ISP is, connection 
performance is influenced by server bandwidth as well. So I guess the 
only reliable factor would be some *real* measure. Which is far from 
being absolute, as the network continuously varies its state. Probably 
you should repeat this step over and over in time to get some average 
valuable value, that you could then use (for example) to reprocess 
images on the fly and have output graphic resolution degrade with 
connection quality to keep speed close to constant (just an idea, not 
that I'll try to do that very soon) :).

Still your test must take the most minimal transmission load ever 
because it would add up and further lower the bandwidth of those who are 
sitting on a poor connection. Maybe you could add it to the banners. It 
could serve both as an "okay, printed" mark *and* a performance 
benchmark. Just have the timestamp recomputed on an invisible iframe (or 
have something sent along within the link the next time you load a 
banner, it depends on how quick you need an answer).

Won't work for *all* browsers, but say 90% of your users will give you 
the data you look for. You'll have a default strategy that appies when 
no answer comes, of course.

For those having banners sent out at regular intervals the average value 
would be reasonably close to the truth, and with little overhead. Of 
course you would need to take off the graphical load (size of the image) 
from your *packet* dimension, when computing the result.

Just my 2 kopeki :)


LoRd, CaN yOu HeAr Me, LiKe I'm HeArInG yOu?
lOrD i'M sHiNiNg...
YoU kNoW I AlMoSt LoSt My MiNd, BuT nOw I'm HoMe AnD fReE
tHe TeSt, YeS iT iS
ThE tEsT, yEs It Is
tHe TeSt, YeS iT iS
ThE tEsT, yEs It Is.......

PHP General Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php

Reply via email to