Re: Embellished operators

The case I was referring to is "an mrow whose arguments consist (in any
order) of one embellished operator and zero or more space-like elements" is
considered an embellished operator. The rules for how this stretches are
not clear because the definition as to what its unstretched "normal' size
are not specified (the consensus is that the size of the space-like
elements should be taken into account for its normal size).  But I think
this case is extremely rare in practice.

The table case is used a lot (ok, some).  The rules for that are (I think)
clear in "3.2.5.8.3 Horizontal Stretching Rules":  "... should stretch to
cover the width of the other direct sub-expressions in the given element
(or in the same table column)...".  Or to specifically answer your question
-- it should stretch to cover the widest of the non-stretchy column
entries, not just the text in the mover (but it needs to stretch to match
that at a minimum).

Hmm -- if you wanted to prevent the mover from stretching to covering all
the other cells, then I'd probably add a zero-width mspace.  But that would
be in an (implied) mrow, and thus considered an embellished operator...
which is exactly what I do NOT want to happen in that case.  I wish I could
remember why that rule was added.

    Neil




On Thu, Mar 29, 2012 at 4:42 AM, Frédéric WANG <fred.wang@free.fr> wrote:

> Yes, that's similar to the horizontal case that I initially mentioned (and
> for which there is a bug in Firefox since the embellished op rules are
> completely implemented). I think Neil means that the vertical case does no
> really come up in practice. In your example, that would be replacing the
> mover by
> something like <mrow><mo>(Vertical Arrow)</mo><mspace
> height="100px"/></mrow>...
>
>
> On 29/03/2012 13:10, Davide P. Cervone wrote:
>
>> Actually, I think it does come up in practice:  in mtables that are used
>> to produce commutative diagrams.  With cells that contain horizontal arrows
>> with labels above using mover, should they stretch to match the label, or
>> the maximum width of the column, or both?  If an arrow has a very wide
>> label, as an embellished operator, is its natural width the width of the
>> unstretched arrow, or the wide label?
>>
>> Davide
>>
>>
>> On Mar 29, 2012, at 1:57 AM, Neil Soiffer wrote:
>>
>>  Sorry for not following up.  I was at a conference during the last
>>> working group meeting and this didn't get discussed.  I'll bring it up and
>>> see if we can get some official position on it at our meeting next week.
>>>
>>> Based on what you and Dave said, I lean towards option 2.  Actually, I
>>> rather remove the rule (which I think Dave wants too), but that's not an
>>> errata and would need to be done in a spec revision.  In any case, I doubt
>>> this comes up in real life and I suspect it is near the bottom of the heap
>>> in importance in bug reports for Firefox, MathPlayer, etc.  Still, it's a
>>> hole in the spec and needs to be filled...
>>>
>>>   Neil
>>>
>>
>