> On Sun, May 06, 2001 at 08:38:18PM -0500, Alan Shields wrote:
>> The interesting part comes in when we consider that the majority of
>> low-bandwidth users will probably be connecting to a higher-bandwidth fproxy
>> (casual users, and such). If the fproxy were to scale down the image
>> transferred to match the markup, then only when the user requested the
>> larger file would it be transmitted over the slower link, and the node would
>> still be prepped with the image.
>
> Well actually fproxy should be running on localhost most of the time
> making the local transfer cost insignificant.
>
What prevalence should we expect of transient, low-bandwidth nodes? These
nodes will mostly be acting as leeches, since the probability of uploading
something from that node, rather than the higher-bandwidth parent node, is
small?
Or am I confused on this issue?
>> This would probably only be easily done in the reference implementation with
>> boxes that have ImageMagick loaded on them, as it would become a simple OS
>> call.
>
> I don't think that the Java->ImageMagick interface is too well developed
> ;-)
>
> Ian.
>
Yes, this would be a quick hack-in on the reference implementation. Mostly
I'm trying to consider what measures we should take to reduce bandwidth for
beginning users, in light of the fact that the average freesite will take up
more space per uncached request than the average website (considering the
power-of-two padding, but more considering the shape difference between a
well-constructed freesite and a well-constructed website).
More so, should we even consider such things in a reference implementation?
Or maybe I should just hit the crack pipe a few more times and call it a
day.
--
"Y'know...I'm not sure why, but there just seems something wrong about
watching porn while eating breakfast."
--Bimbo, from Exploitation Now (Poe)
_______________________________________________
Devl mailing list
Devl at freenetproject.org
http://lists.freenetproject.org/mailman/listinfo/devl
