Paul B. Gallagher wrote:
Richard Owlett wrote:
Martin Feitag wrote:
Paul B. Gallagher schrieb:
With modern broadband connections, caching doesn't really speed up page
loading that much anymore the way it did in the dial-up days.
If you define "broadband" like the German government, it's still an
important part of the browser for some people. ;-)
Last time I checked, less than a year ago, approximately 40% of US
users ( including me ;) are still on dialup.
It's hard to judge, because "broadband penetration" statistics divide
the number of broadband access points by the total population, not by
the number of Internet users. US stats for December 2008 were around 25%
by that measure, but how many users had access to each line? And what
proportion of the population isn't online at all?
As for speeds, OECD includes anything over 256 kbit/s, which is a pretty
low cutoff but does exclude dial-ups.
More info here:
Interesting links. Motivation to see if I can find again the
article I remembered. Were measurement methods different or was
the article more dated than I thought.
There's a subtle trap for software developers - assuming that their
setup being 'nothing special' is not only 'normal' _BUT_ *THE NORM*
I wouldn't call it so much a trap for SW developers -- they don't suffer
if they get it wrong, we do. Typically they write for high-end systems,
driving innovation as users demand better performance from manufacturers
and ISPs. A minority will write for low-end systems in the interest of
broader market penetration.
I think the traps more subtle. For example, top of line systems
are so big and/or fast that the problem may not be observed. Is
all this bandwidth needed or is does it cover up other problems?
support-seamonkey mailing list