lynx-dev
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: LYNX-DEV Optimizing relative URLs


From: Al Gilman
Subject: Re: LYNX-DEV Optimizing relative URLs
Date: Sun, 3 Nov 1996 11:03:13 -0500 (EST)

  From: address@hidden (David Woolley)
  
  Al Gilman wrote:
  > 
  > OK, so you are referring to those who run Lynx on their own Linux
  > or Un*x box with a dialup PPP connection.  That was not what I
  > imagined when you said "dialup user."  I thought "dialup user"
  > means someone like me who dials into a shell and runs Lynx there
  > -- but these users can't take this DNS caching dodge.
  
  Any competent shell account ISP is already running a caching nameserver
  on the machine you dial into.  If they are not, you need to encourage them
  to do so.

I debated whether to answer you or not, as this thread is
unlikely to change anything from here on in.  Maybe we can find a
way to close this thread by asking what the possible
actions/outcomes are that pertain to the interests of this list.

1) Using IP numbers for URLs pointing to the lynx-dev archives
from within Lynx public information pages.

        Do you think this is OK?  I am willing to update the FAQ references
        on the unlikely chance that U. Kansas changes IP numbers for
        the archive without changing the node name.

2) Lynx caching IP numbers with its history of accesses:

Here the expected performance of ISP name service matters.  I
seem to have different expectations of ISP-provided name service
performance from the expectations you hold.  I expect that ISPs
will provide only enough name-lookup speed [from their, caching
server] to keep Netscape/PPP users from complaining.  Since Lynx
uses more name-lookup in proportion to HTTP data transfer
(because of images not transferred, and generally switching pages
more often) than Netscape/PPP uses I expect that the equilibrium
speed of ISP support to name-lookup will leave the Lynx user
wanting more.  The average Lynx user is likely to be in a
position to gain from the optimization provided by IP cacheing
internal to the Lynx process memory.  I expect that under
market-equilibrium performance from commercial ISPs that the Lynx
user will get noticeably better service if Lynx caches the IP
number than if Lynx doesn't.

This doesn't make IP caching in Lynx urgent.  But it does mean
the opportunity to do better is genuine.  The opportunity isn't
going to go away as ISPs get "competent."  The ISPs that survive
are only those that remain slavishly bottom-line driven, and Lynx
doesn't drive much of their bottom line.

Al Gilman
;
; To UNSUBSCRIBE:  Send a mail message to address@hidden
;                  with "unsubscribe lynx-dev" (without the
;                  quotation marks) on a line by itself.
;



reply via email to

[Prev in Thread] Current Thread [Next in Thread]