Re: lynx-dev <BR>

[David Woolley]

> 
> Note that runs of   's apparently should also be collapsed,
> but lynx *doesn't* Do The Right Thing in that case.

  is not white space for the purposes of section 9.1 of the 
18 Dec 1997 HTML 4.0 specfication, although there is a general warning
about not relying on the handling of spacing characters other than those
defined as white space.  (Use style sheets instead.)  In fact, the whole
point of   is that it not be treated as white space, but rather as
a printable character with no pixels painted.

With regard to mixing of white space characters, I think the intention is
that they should be collapsed to a hypothetical white space character and
that that should be rendered in a language dependent way.  In particular,
I would expect space, tab and newline to be treated alike, although might
expect a sequence containing only zero width spaces to be treated as a hint to
generate a zero width space.

I think that the description of BR is confused because it is in the same 
section that defines newline as white space.  The browsercaps site doesn't
actually specify HTML 4, and it is possible that earlier specs are clearer.
I don't, though, think that BR was intended to be treated as white space.
(Note that there is an explicit requirement to collapse redundant Ps.)

BR causes problems for style sheets as illustrated by the comment in the
sample style sheet in the CSS2 spec and the lack of any style definition
for BR.  Although BR is not declared deprecated, it seems to me that DIV
or P with an appropriate style might give rather more predictable formatting.
(Note Lynx doesn't support styles.)