Re: Accents in the operator dictionary

Frédéric WANG writes:

> 1) and 2) are actually related. I was wondering if the rules to
> determine the form are enough to choose the accent in case 2),
> even when there are multiple forms (there was the same problem
> with "tilde", but apparently there are now two distinct operators
> "tilde" and "tilde operator"). My answer to 1) was that these
> accent operators are used in munder/mover/munderover after the
> operand (the base of munder/mover/munderover) so they are
> postfix. I try to justify it by the rules below, but I failed:
>
> http://www.w3.org/TR/MathML3/chapter3.html#presm.formdefval
>
> For low line, I don't see what is this infix operator but I was
> not asking to remove it from the dictionary. Rather than dealing
> with each particular case I would prefer a general solution. For
> Mozilla, if accents are really postfix, I was thinking on adding
> the rule above when the operator is a child of
> munder/mover/munderover other than the base (why do you think
> about that, Karl?)... However, the spec does not seem to allow
> implementers to define their own heuristic rules.

Yes, the spec gives implementations flexibility in operator
dictionary entries but less flexibility in when to use them.

Implementations do however have responsibility (even if not
flexibility) to decide in which directions an operator gets
stretched.

"It is up to the renderer to know in which directions it is
reasonable to stretch a character"
http://monet.nag.co.uk/~dpc/draft-spec/chapter3.html#presm.op.stretch

Would it be reasonable to make infix operator "_" default stretchy
and stretch by default in all horizontal stretching situations but
never in vertical stretching situations?

Certainly with "_", vertical stretching is not reasonable.  I
don't know whether there are any characters that would normally
stretch horizontally as accents (default stretchy) but only
vertically when there is explicit stretchy="true" (default not
stretchy).

Received on Friday, 23 April 2010 00:57:36 UTC