Re: <br> as paragraph separator

On Wed, Dec 15, 2010 at 11:30 PM, Aharon (Vladimir) Lanin <aharon@google.com
> wrote:

> > In IE7, and IE8 and IE9 Beta in quirks mode, <br>
> > is treated as a paragraph separator. However, in
> > IE8 and IE9 in standards mode, it is treated as a
> > line separator and does not end the bidi paragraph.
>
> Wow! I had not noticed this.
>
> I guess standards really means standards, which is good. And they can
> always fall back on quirks and "strict" mode, to keep existing documents
> from breaking.
>
> The question is whether this means that the change re <br> in HTML5 should
> be rolled back.
>
> My gut feeling is it should not. Regardless of what <br> is supposed to be,
> authors usually assume it is the equivalent of a newline in plain text, and
> bidi users (but not monodirectional users!) are very unpleasantly surprised
> when it does not behave the same way.
>

What newline does in plain text depends on the application and operating
system, and is often surprising by itself. The expectations of MS Word users
are different than those of GNU Emacs users, etc. Any way you do it, there
will be some people more surprised than others, and I have no clue exactly
which audience you should target.


> Using <p> and </p> is challenging in many scenarios, and so the use of <br>
> where <p> really should have been used is rampant. As a result, it is that
> much more difficult to work bidi support into web apps, since they must be
> changed to stop using <br> as a paragraph separator. This request is often
> treated as unreasonable and never implemented.
>

If the standard is clear enough, and if web authors want to write HTML5
compliant sites, there should be no problem with their new sites.
The bidi problems would always remain for HTML4 sites, but there is nothing
HTML5 can do about it.

There is no point in forcing the standard to follow some behavior only
because it is a strong market trend.
If you do that too often, there would be no incentive for people to follow
standards. Why spend time on making your site HTML5 compliant, when if you
follow the strongest market trend instead, it would probably BECOME the
standard in HTML6?


> Thus, IMO, we should leave HTML5 in its present state, i.e. with <br> as a
> paragraph separator. Since standards seem to be being followed in standards
> mode, I hope that IE9 will revert to the old behavior.
>
> However, I would be very interested in hearing others' opnions, in both
> directions.
>
>
As I already said, I do not like the current situation. At the current state
of things, my opinion is that it would be best if <br> is deprecated and, in
due time, completely removed from HTML.

This can be done gradually: at first stage (HTML5), the standard should
MANDATE implementation of unicode entities for (both) LS and PS in any
visual browser - these should be used by CSS (or other display mechanism) to
control the flow of lines. In addition, at the description of <br>, a remark
should be added that this is a feature kept for backwards compatibility and
easing migration, and that <br> is not recommended in new sites (use either
<p> or the LS/PS entities instead).
The next release (5.9? 6.0?) would completely remove it.
If this suggestion is accepted - I think (not sure though) that the current
fix should be reverted (this is in light of Simon Montagu's recent findings
- backwards compatibility is a tricky thing...).

Since my time is limited right now, I intend to try pushing this suggestion
via the bugzilla system in January - unless you people convince me
otherwise.


> > Even more surprisingly, in IE8 in standards mode,
> > a line break in preformatted text is also treated as
> > a line separator.
>
> Bug.
>
> > IE9 Beta has reverted to treating it as a paragraph
> > separator.
>
>
Is that in standards mode or in quirks?


> Good.
>
> > I also discovered strange behaviour in Safari 5.0.3:
> > if a <br> or line break in preformatted text is followed
> > by a rtl embedding, e.g.
> > <div>ltr RTL,<br><span dir="rtl">MORE RTL</span></div>
> > the comma at the end of the first line is rendered to the
> > left of the rtl text, as if there were another rtl character
> > following it.
>
> Very strange. Also happens in Chrome. Looks like a WebKit bug.
>
>
Apologies for being more argumentative than informative (theoretical
arguments are less time consuming than actually checking up stuff :-) )

   AA