Arrows in menclose needed?

Hi all;

Recently Frederic Wang was working on implementing the \cancelto
macro and proposed an local extension (for Gecko & MathJax)
of MathML's menclose notation to support it.

Recall that \cancelto (in the LaTeX cancels package) draws an arrow
from the lower-left to the upper-right, with an arrow head at the
right end, as it were pointing at the superscript.

Recall also that MathML's menclose allows the author to specify
from an open-ended set of notations, with standard ones including
updiagonalstrike and downdiagonalstrike. No arrows are predefined,
but using notations of your own is perfectly valid.
For historical reasons, updiagonalstrike is defined to be
"from lower-left to upper right"; not necessarily the most RTL friendly.

Frederic's proposal is as natural as can be, completely in the spirit
of the MathML spec:
  <menclose notation="updiagonalstrike updiagonalarrow">...
Agents that support the "updiagonalarrow" notation will show an arrow,
but others will show at least a diagonal strike-through without the head.

Analogously, downdiagonalarrow could be proposed, and likewise
verticalarrow and horizontalarrow could be, but then a problem
becomes apparent.

The problem --- if indeed there is one --- is that this obvious naming
   "updiagonalarrow"
doesn't easily extend to putting the arrow head on the other end
(if we'll ever need to?), nor to adapting to RTL (Does RTL use arrows?),
or other kinds of arrows, or...

So, I'm wanting to sound out the community, both the authors and
implementors on what is needed and sufficient here. The simplest
solution that fulfills current needs without poisoning future
needs would be ideal.

Is updiagonalarrow the only arrow needed? Are other directions needed?
Is the spec fine the way it is, without arrows?

Let's be honest: if Gecko and MathJax adopt a convention,
it may as well be in the standard...

It is perhaps remarkable that MathML has been in play for ~15 years
without a clamor for arrows. But the huge range of exotic arrows
that _are_ apparently needed for some applications and which are
supported by Unicode is also remarkable.

Sorry to be so wordy, but the context of the question
is hard to state concisely.

Thanks for your attention
Bruce Miller,
mostly on behalf of the Math Working Group

Received on Tuesday, 18 June 2013 23:01:40 UTC