Re: lynx-dev alarm: Lynx will be soon an old unusable browser

[Martin McCormick]

        I am one of the lynx users who would love to see lynx
work on sites using relative links and, of course, javascript.

        I have been experimenting to see what appears to break
when using lynx with a site created with Lotus Domino as that
seems to be a textbook case of everything that is wrong with
vendor-driven technology.

        The trick, here, is to fix this problem without breaking
something else.

        I have compiled lynx2-8-5 with the -g flag set so I can
snoop on it's operation by using gdb.  I have a lot yet to learn,
to say the least.

        One experiment I did, however, has already tought me
something which I will share.

        I used the snort utility in Linux to see exactly what
gets sent when one hits enter to select a link in the lynx
browser.

        Lynx sends a get command followed by the url one is on.
When I reach a link that is relative and broken as far as lynx is
concerned, lynx sends a get command that is corrupted by some of
the html that was in that link.  I was expecting to see either
nothing much or a path that doesn't work in relation to the
current root directory.

        What I see, however, is not a path anymore, but part of a
path plus some garbage.  What is amazing is that lynx and the
Lotus server still keep trying to communicate.

        I hope somebody does beat me to a solution to this
problem because I am having to start from square 1 and I know
there are those who have been building browsers and servers for
years who can get there a lot faster.  The problem with relative
links seems to be that the lynx browser can't distinguish them
from good or absolute links.

        Javascript is important all right, but the lynx browser
would probably do well on many sites if it could only follow
relative links without getting confused.

        The more I learn about how the lynx browser works, the
more I think it is basically very sound.  It needs to stay that
way plus have an engine to convert what it presently does not
understand in to what it does.  After all, it is text that is
sacred and all we need to do is to be able to either dig it out
of the markup language or produce a working link to the binary
such as a pdf or Heaven-forbid, an image file that can be
run through an OCR application.

        As I see it, there are two lynx killers on the loose
today.  The easiest dragon to sleigh is probably the absolute
versus relative link issue.  The other beast is the javascript
components that deal with text processing or downloading of
binary files such as sound that one might use if not for the
navigation problems.

Martin McCormick WB5AGZ  Stillwater, OK 
OSU Center for Computing and Information Services Network Operations Group

; To UNSUBSCRIBE: Send "unsubscribe lynx-dev" to address@hidden