Help! Links w/Lynx don't work

[Thomas Dowling]

[Quoting Andrea Duda]
> >
> >...Apparently Netscape can handle URLs that are split across two
lines, but 
> >Lynx cannot.  When I edited the page so that URLs were on a single
line, 
> >Lynx was able to connect to all the search engines without a
problem...
> 

This suggests to me that Lynx is being somehow deficient in handling
URLs with white space in them.  The URL specification uses the term
"unsafe" for difficult characters like a space, rather than "illegal"
or "forbidden".  The spec says that all unsafe characters must be
encoded, but it doesn't offer any guidance on how to handle URLs that
include them anyway.  Since most host operating systems are capable of
managing files with spaces in their names, it's quite conceivable that
an author really would write a space into a URL (quick, who knows the
hex character triplet for a space?), and that a text editor would save
that space across a line break.  I'd rather my browser gave authors
credit for saying what they mean--even if they have problems with the
syntax--than to waste time and lines of code trying to guess what they
really meant.

[TPD gets off soapbox momentarily and notices that all the extraneous
line breaks in Joe's URLs were inserted into domain names, not file
names.  Well, that's a horse of a different color, but I could still
sit around all day finding higher priorities for the gang at Microsoft
and Netscape to work on.]

[Quoting Joe Schallan]
> ...Carole Leita made the wise suggestion to run all HTML through a
> validator.  I hadn't taken the time to do that, of course.  Good
> idea.

A strict validator would not have detected the problem, since URLs with
white space apparently don't violate the DTD--the HTML spec and the URL
spec don't always see eye to eye.  The WebTechs validator, for example,
has comments about unquoted font size attributes, an extraneous body
element, and a misformatted comment.  In fact, the document also passes
the Spyglass validator and the Weblint and Doctor HTML checkers.  The
only thing I could find which pointed out the problem was htmlchek, a
checker that hasn't been updated in two years and which I had almost
deleted as being obsolete.

Just goes to show that there's no substitute for casting an eye over
the source code.  


Thomas Dowling
Ohio Library and Information Network
tdowling at ohiolink.edu