[Om3] three avenues for the condition elements

[Professor James Davenport]

On Sat, October 25, 2008 5:17 am, Michael Kohlhase wrote:
> the MathML WG has to bring out the next working draft of the MathML3
recommendation and we have a code freeze on November 6. Since this is
the last working draft before the "last call" stage of MathML3, it would
be good to have resolved the question whether the <condition> element
belongs into strict MathML or pragmatic. Therefore I would ask you to
give your opinions about the resolution ASAP. I will organize a
> teleconfere later in the coming week or early in the week of the 6.
where we take a decision.
A good idea. I apologise for the delay in not following up on the teleon
of 10th - as usual I was overtaken by teaching before I had a chance to
finish the follow-up - I attach the current state for what it's worth.
[second thoughts - it wil probably fall foul of the length limit: I attach
the LaTeX, and the PDF (which may be further updated if I get a chance
today), is at http://staff.bath.ac.uk/masjhd/Conditions-JHD.pdf]
> Paul Libbrecht wrote:
>> (warning: this text contains unicode character)
>> #1 condition element
>> Basically add, in OpenMath and strict MathML, an element called
condition or omcond that mimics the current condition element.
>> #2 condition symbol
>> Invent a new symbol called condition which would do a very similar
function.
>> For example, if it was called c, one would write a conditional
>> function as
>>   λ.x,y: c(x ≠ y, x / (x-y) )
My fundamental objection to MathML's condition, which I think applies to
both #1 and #2, is that its presence affects the semantics of the
surrounding objects. Indeed, I do not even no of a formal statement about
how far its influence might spread: it is sufficient to inspect all the
direct children of X for a condition element, or must one delve deeper?
>> #3 conditional symbol variants
>> For each binder-like symbol, add a variant symbol which does accept an
extra argument, the condition. The function above would be written:
>>   λ.x,y: x ≠ y, x / (x-y)
To do this neatly, one would have to scrap the rule that OMBIND only takes
one 'ordinary' child, but this, I think, is a small price to pay. Indeed,
if it is felt to be too high, one could add OMBINDCOND, with two
'prdinary' children, the first as in OMBIND and the second the condition.
This would rescue MK's DEFINTCOND from the objections in my attached note.

James Davenport
Hebron & Medlock Professor of Information Technology
Formerly RAE Coordinator and Undergraduate Director of Studies, CS Dept
Lecturer on CM30070, 30078, 50209, 50123, 50199
Chairman, Powerful Computing WP, University of Bath
OpenMath Content Dictionary Editor
IMU Committee on Electronic Information and Communication


-------------- next part --------------
A non-text attachment was scrubbed...
Name: Conditions-JHD.tex
Type: application/x-tex
Size: 11927 bytes
Desc: not available
Url : http://openmath.org/pipermail/om3/attachments/20081026/999977e9/attachment.bin