Re: [MathML3-PR] Alignment with display="block"

A.M.:

>>While looking into what a user agent such as WebKit should do by
>>default for when math elements have display="block", we've read
>>through Section 2.2.1 where it says:
>> . . .
>>There is a great preference out there for block level Mathematics
>>to be centered.  The second preference would be for it to be
>>at least indented from the normal block flow.

Yes, centering is preferred for displayed math.  Note that the general
concept of block in HTML includes things like "blockquote" and the
various list structures where a set-off from the margin is the
preferred visual rendering.

>>Is it the intention of MathML3 for block level Mathematics to default
>>to left or writing direction start alignment?

I hope not.

>>If so, that would require users to add additional rules to their CSS
>>to change to centered or indented block Mathematics.

I think one should say 'authors' rather than 'users' here.

It's certainly a good idea for the author's preference to be given in
CSS.  (I believe there are some browsers that do not automatically
center.  Also I believe there are some browsers that do not support
CSS inside the math namespace though it is supported for the html
namespace.)

S.D.:

> It seems to me a user agent has leeway to supply default style rules
> that support the preference for centered/indented block math elements,
> despite the general default for other block elements, in the same way
> legacy HTML attributes are given behaviors in terms of CSS properties.
>
> I think the intent of 2.2.1 is simply to recommend that the user agent
> take the display attribute into account when it produces a rendering,
> without over-constraining what that rendering should look like.
>
> So if a user agent centers the block math elements by default, and if
> users must supply style rules to achieve other behaviors, that for me
> would be one possible conforming behavior.

Yes.

N.B. Firefox 3.6.8 centers math with display="block" by default.
As I recall, it and the pre-Firefox math-enabled Mozillae always
have.  (In early days the switch was, I think, mode="display".)

While I'm here, I'll express my opinion that math displays should by
default never wrap but go into "overscan".  I see it the business of
the author to manage lines in displayed math.  So the CSS
"white-space: nowrap" should be in effect by default.  Authors who
want wrapping should be able to employ "white-space: normal".

                                    -- Bill

Received on Thursday, 2 September 2010 16:12:15 UTC