Re: (not so)Quick Update on my progress

[Gregory Maxwell <nullc@nightshade.ml.org>]

On Thu, 26 Jun 1997, Kultarr wrote:

> ---Gregory Maxwell <nullc@nightshade.ml.org> wrote:
> [He's been busy with the cache problem.]
> 
> A couple of points...
> 
> 1) Good work.  Has anyone discussed the ability to selectively
> download and cache graphics from pages?  Over a modem, I have
> often used lynx for browsing because it is blazing fast by
> comparison to browse for information and not have to wait for lots
> of graphics.  [Netscape used to render the HTML onscreen early
> and then backfill the graphics portion, which was another
> (and good, IMHO) approach.]  I guess applets would also fall
> under something that could be selectively deferred.
Thanks. Yes, I've thought of that.. I've got a big section on how the
cache integrates with the rest of the browser.. but hush.. your not
supposed to know about that yet.. :) I'm kinda enjoying listening to the
squid people tell me "There is no reason to implment the cache inside the
browser on platforms where they can run an expernal cache".. I plan on
showin 'em..

> 
> Basically, I guess I prefer the "display as much information
> as soon as you possibly can" approach.
Thats what I would like too...

> I guess this is more of a display manager topic than cache...
The cache can help with that..

> 2) You mentioned snooping on other "local" caches... I had a
> couple questions on that aspect.  How do you know what other
> machines "are close" to the current machine in general?  What
> about security/privacy issues?
Right now I detect them by sending and ICP ping (a udp packet) to
networkaddress.255.. Then I would see if there are any responces with a
reasonable 'ping time'.. Then I would prompt the user.. (

 > 
> 3) I like the idea of prefetching small HTML pages.  Another 
> enhancement could be to put some predictive intelligence (based
> on history and content) to determine where the user is likely
> to want to go next.
Yes,yes yes... Already in the works... (I index the title lines of pages
and try to pattern match..) :) I.e. if you are on a page called "hot dogs"
and the last page you were on was called "baseball" and the page before
that was called "sports" It would first grab links with the words "hot
dogs" then "baseball" then "sports".. hehe.. I've actually tested this out
so far... When downloading the html on pages <8links... It gets over 90%
hit rate.. And 35% first time (meaining it picked the right one..  but I
cheat.. I also download from top to bottom.. which is how people browse)..


> 
> ===
> Kultarr
> 
Gregory