Re: Math on the web without MathML (CSS 2.1 rendering for HTML and XML)

juanrgonzaleza@canonicalscience.com wrote:
>
> Mark P. Line wrote:
>>
>>> Successful means winning, doing well, triumphant...
>>
>> Like Microsoft Windows, then. Okay. That corresponds fairly closely to
>> my meaning of 'successful' here.
>
> Except that Microsoft Windows does not work for those tasks better done
> with several Unix, Linux, real-time OS...
>
> Maybe you are now confounding "successful" with "popular" or
> "mainstream".

No.

Actually, I belong to that particular community of practice whose job it
is to explain to others exactly what English words mean, so it would be
exceedingly unlikely for me to have confused those words. You may not have
known that, of course, so your use of this rhetorical tactic may be
understandable.


>> I think MathML works, and I think it's successful in the areas it's
>> useful
>> for. That success is demonstrable in terms of market penetration, just
>> like the success of Windows.
>
> I (and others) think that MathML is not successful and is not working in
> the way has been popularized. Enough examples on why MathML code is being
> spreaded on the web is often poor than old alternatives was extensively
> discussed at this list. No need for addressing your "market penetration"
> again.

It all boils down to our difference in opinion about what it means for
MathML to be considered "successful". Let's agree to disagree. That way,
you can allow words to mean whatever you'd like them to mean.


>>> Fortunately, not all people has so great views on MathML like you; just
>>> some thougts from the SBML community:
>>
>> My views of MathML are not that favorable. If you look back at what I
>> wrote originally, you'll see that I said that I use MathML because it
>> *IS* the standard and that, if left to my own devices, I could have
>> easily created a language that would suit my purposes better than
>> MathML2 (but not necessarily the purposes of others at the same time --
>> there's the rub).
>
> Again weak semi-defined arguments. What exactly would I understand by
> your "it *IS* the standard"?

There are at least two ways to understand what I meant by that:

(a) The W3C has exactly one standard for representing math, and that's
MathML. Like you, I choose to orient myself largely around the W3C (as
opposed to ANSI or ISO or whatever). There's a particular reason for that
in my personal case: the W3C gets stuff done in finite time. I was a
consultant to the original SGML committee back when the earth was still
cooling, and I argued fruitlessly for decades to get a standard,
stripped-down but still SGMLesque language for hierarchical data
representation and transfer. It was the W3C that ultimately caused XML to
be successful, and it will be the W3C that will ultimately cause most
current and future IT standards to be successful.

(b) The computational biology community that uses CellML and SBML (two
standards for representing physiological and metabolic networks) has
adopted MathML as their standard for representing the mathematical bits
(e.g. rate laws in reaction kinetics).


> Any case ISO-12083 *IS* an previous international standard (MathML is
> technically a w3c recommendation not a standard because w3c is not a
> standards body) and I do not see you promiting its usage, why? What are
> your reasons?

Because I don't like ISO, and do not wish to place any bets on them. This
is a business decision on my part, you see -- not an academic decision,
much less an aesthetic one.


>> It's not surprising that people in the SBML community would say the same
>> thing. Great minds think alike, after all.
>
> Let us wait and see if the introduction of MathML into SBML was fortunate
> or not :-) Do not forget that great minds can make mistakes too.

We already know that it was fortunate. I can build a tool to plug numbers
into content MathML expressions that will work with both SBML and CellML,
and with anything else that uses MathML. I can build a tool to turn
MathML-expressed systems of differential equations into a discrete event
graph for simulation without needing to come up with my own ad hoc
language for representing the equations.



>>> And MathML is so succesfull in the SBML community that last years was
>>> launched a thread in the SBML site called "Complementary Alternative to
>>> MathML Needed". I do not believe that mathematics done at the SBML
>>> community was extensive or advanced still they are claiming for
>>> alternatives.
>>
>> I think you need to look up "complementary" in your favorite English
>> dictionary.
>
> [...]
>
>> I'm in fairly good company when I define success of a product (even a
>> public, community-created product like a W3C REC) in terms of market
>> penetration: how many sites are using it vs. how many sites are using an
>> alternative.
>
> Well, in that case MathML and OpenMath are not penetrating much. Yes
> there is some usage of MathML off-line at the publishers side, but usage
> of MathML online is minimal.

Even if it's true that usage of MathML online is minimal, so what? Why are
online applications more important than others? You can't say that MathML
is unsuccessful across the board just because it's not found as much use
in your favorite horizontal than in others.

(Well, actually, *you* can say that, and often do. But most reasonable
people would not.)


>> Success of a product doesn't mean that I like it particularly well, that
>> I don't think it could or should be improved, or that there are no
>> conceivable alternatives that would better serve my purposes. My
>> opinions about a product have absolutely nothing to do with objective
>> measures of that product's success. It's those objective measures about
>> MathML that I think you're in denial about.
>
> The only problem with this piece is in your "objective measures".

What problem is that?


> A thing irritating me is the only-way concept you (and others) are trying
> to promote without real rationale for leaving alternatives.

When have I ever tried to promote any kind of only-way concept with
respect to MathML? I'm sure you must have misunderstood me.


> How expressed many times here if MathML is so good and so popular and so
> succesfull, nobody would become interested in alternatives, therefore you
> would not worry about my writtings after all.

I worry about your writings because you do not take a careful and reasoned
approach to making your points, because your writings are very difficult
to understand completely and because you spend a lot of bandwidth fishing
for debating points with clever nonsequitur or decontextualized flippancy.

But I worry about global warming and peace in the Middle East, too, and
there's not all that much that I can do about those problems, either.


>>>> What do you mean by "content" and "meaning" such that content MathML
>>>> is
>>>> about the one but not about the other? I don't think these words mean
>>>> what you think they mean.
>>>
>>> Then you already know the reply better than me.
>>
>> If I did, I wouldn't have asked. So your answer is that you don't know
>> what difference there is between "content" and "meaning" with respect to
>> content MatML. That's why I don't understand why you think content
>> MathML is about "content" but not about "meaning". But never mind,
>> you've already answered as best you can.
>
> You see! You already knew better.

Not at all. I continue to call them like I see them, which is the best I
can do.


>>> Wait, first you claim that MathML is sucesfull and apparently your
>>> basic
>>> evidence is because MathML is included in SBML or something but now you
>>> cannot encode those simple examples.
>>
>> "Or something"? Have I been unclear about what I think about MathML and
> why?
>
> Well your "objective measurements" do not convince some people (I am not
> the only one)

Why should I be concerned that the objective measurements I'm aware of do
not convince some people?

There are plenty of people who will remain unconvinced no matter how much
evidence of whatever quality is placed at their feet. Although I feel
sorry for these people, I can't take the time to bring them on board at
any cost. They'll have to catch up later.


>>> Time? Do not you have time enough for encoding examples 1), 2), or 3)
>>> in
>>> content MathML; that sound rare.
>>>
>>> Well, I can provide you a billing address if that is your only problem.
>>> Could you encode above examples for this week please?
>>
>> I was being facetious.
>
> My offer for billing you for the encoding continues :-)

I'm sure I don't want to be billed for doing work for you. I generally do
things the other way around.


>> If you don't know how to encode your examples,
>
> Well I think that I have some ideas about how do it, but am anxiously
> waiting your reply.

As I said, I'm deferring to those who have much more encoding experience
than I.


>> there are lots of people
>> here who would be much better able to teach you than I.
>
> Maybe but reply was addresed to you. You claims MathML is succesfull in
> "market terms" and you are promoting it without solid technical basis.

When have I ever promoted MathML in any way?

All I've done is disagree with you when you've claimed incorrectly that
MathML is unsuccessful in every way (or equally incorrectly, in all ways
but one).

Never in my life have I ever suggested to anybody that they should use
MathML instead of some alternative. My decisions to use MathML are my own,
but you don't seem to want to leave me to them: you tell me I've made the
wrong decision because you think I don't have the technical wherewithal to
make the correct decision, which in your mind must always necessarily
exclude MathML.

Right?


> You also find CSS rendering of no interest

I find CSS rendering of no interest *to me*. What's wrong with that?


> and even initially claimed that rendering of math would be addressed in
> another part

Another part of what? What do you think I initially claimed?


> (even claimed that math on the web posts would be submitted outside this
> list!)

I claimed that it was rhetorically ineffective to post anti-MathML polemic
on this MathML-aligned list. And so it is.

I did make the suggestion that your posts might be better suited to the
WebMath list, but it turned out that the owner of that list doesn't want
us there either.


> You also claim that you are using MathML because is the "standard".

Yes. What should I use for embedded math in CellML and SBML models instead
of MathML, in your opinion?


> After all one of advantages of using a "standard" is many people can
> write and interchange code in a unified way, no?

Yes.

> How is possible that you can maintain that MathML is working and is
> succesfull in function of "objective data" but cannot even encode
> something so simple as (a+b) in content or parallel MathML.

What makes you think I cannot encode (a+b) in MathML? I choose not to
allow myself to be patronized, but that's a different matter altogether.


> Are your claims about success and working so political and marketing
> based as youir thoughts about Windows or Beta or there is real stuff
> beyond your thoughts?

I have to work for a living, so politics and marketing are "real stuff"
for me. Academic and aesthetic vagaries, on the other hand, don't pay the
bills. YMMV.


> I know claims from other people (including MathML folks); i agree with
> some and disagree with others but you sound more as waporware to me.

If you think my work is vaporware, I don't have a problem with that.

BTW, how many employees are there at the Center for Canonical Science? Is
it all housed in one building, or do you have a proper research campus?


> It is time also for a MathML 3 spec, I have proposed some basic ideas for
> the new spec (I do not know if will be accepted or rejected). You often
> claim how good MathML would be if you had done!

No.

I claimed that MathML would have been much better *FOR MY PURPOSES* if I
had designed it myself. I also said that that would precisely *NOT* entail
that MathML would necessarily be any better for anybody else's purposes,
and that that's the rub. The WG's task was to come up with something that
would do reasonably well for a reasonable variety of applications, and
it's obvious (from the objective measures you deny) that they've achieved
that.


> If I remember correctly Patrick encouraged you to post here some ideas
> for MathML 3 discussion but your messages continue lacking technical
> stuff in an alarming way.

As I told Patrick, I will provide MathML-related feedback from the
computational biology arena as I encounter or create it. For now, I'm in
the process of pouring the content MathML specs into a Python expression
cruncher so that I have it in place when I'm ready to provide CellML and
SBML interfaces to my simulation engine. For all of that, we're talking
late this year and into next year, not this week and next.

Most of my work with the simulation platform at the moment is actually in
cognitive science (computational linguistics), which is where I have the
most enthusiastic collaborators.



>> If you have examples that cannot be encoded correctly in content MathML
>> by anybody because the language is inadequate, then I'm certainly not
>> the person you need to talk to.
>
> I agree with you.

That's something, at least.


> I was waiting to know a bit more about your "objective
> measures"...

They're not mine, and they've already been posted here. We never repeat
anything, remember?


>>> 2) You appears to unknow that rendering is a critical issue in
>>> mathematics and that design of content MathML markup is related to how
>>> sucesfull the rendering can be.
>>
>> I think that the careful and exhaustive distinction between notation and
>> meaning is a fundamental necessity in all but the most trivial uses of
>> mathematics. I disagree that content (meaning) markup of math must be
>> related to rendering issues.
>
> And again you fail to understand others are really saying.

Just because I disagree doesn't mean I don't understand.


> <blockquote>
> Historically, it's a common mistake to develop markup systems without
> giving much thought to presentation. For example, only when SGML was
> done did one start efforts to create a style sheet language for it
> (FOSI/DSSSL).
> </blockquote>

Yes, because of the traditional definition of "markup" system. That
traditional definition no longer applies, however.

XML is a markup system.

RDF and OWL can be expressed in XML.

RDF and OWL files therefore constitute "markup".

There's no prima facie need to care about presentation issues for RDF and
OWL files.

Nobody think's it's a mistake (much less a common mistake) to have
developed RDF and OWL without giving much thought to presentation.

Therefore, your blockquote is misguided, misstated or misquoted.


>>> It is not how you think that one can design a content markup for people
>>> who is not interested in rendering issues and next one design a
>>> presentational layer for rendering.
>>
>> How do you know? Have you ever tried to design a content markup scheme
>> for math without regard for of rendering issues?
>
> After many posts I can begin to see your point clearly now ;-)

Perhaps. If you're starting to see that I think content MathML has more in
common with RDF and OWL than with, say, HTML or Postscript, then you might
be right.


-- Mark

Mark P. Line
Polymathix
San Antonio, TX

Received on Tuesday, 25 July 2006 18:46:25 UTC