Re: Profiling and certificates for MathML. Avoiding imitators

> > It makes sense to have something similar in MathML, but we don't have well
> > defined profiles yet, in case of MathML 2.0 there is one MathML DTD that
> > allows to mix everything. Also putting two extra attributes on every math
> > element is probably redundant, maybe it is better to specify it once on
> > document level?
> 
> In that way, there is no posibility for mixed documents from different
> authors to be properly merged.

By converting them in single format used by publisher ;) Well, having attribute(s) on math element is the better solution if you need to specify content of each formulae and make sure that information is not lost during copy/paste or mixing. But putting extra attribute(s) on each math element could be an issue for authors and extra verbosity might encourage them not to use these attribute(s) at all. One can assume that version and/or profile attribute(s) of math element are inherited by subsequent math elements if those lack these attributes. In this way one can avoid redundancy in documents with uniform content and still be able to combine different versions in one document, but then again version/profile info will not survive simple copy/paste operations (also applying proflie specific style sheets is a problem as there are no subsequent element selectors in CSS).

> 2007 Practical case b:
> 
> Body B generates tool L. Tool generates valid nice MathML. Market is using
> browser F claiming MathML support for HTML5. You begin distributing some
> units of L, but after some time users reclaim you because files generated
> by L are not rendered by F. It is clear -to users- that tool L is flawed.
> B accusses F from misleading publicity but the damage was done.
> 
> Today, any tool generating <none/>, <plus/> or <mtext/> will be incorrect
> from the point of view of a HTML 5 average user. Would we rewrite those
> 100 tools? What will become next? And what if HTML5 is not so popular as
> some predicted? A new 180º change for the 2012? What then? The *non-XML*
> and *non-HTML* language that certain organization is trying to popularise?
> 
> Some control over is or not is MathML appears to be needed.

But who is supposed to control it? Ideally it should be responsibility of parties involved in process to agree on common principles and then follow them, if it is not the case then it is hard to ensure interoperability just by pointing to papers. For instance if decision was made to treat something as XML application, then individual participants should not undermine the rest by accepting non-wellformed content. In case of SVG for example, there is some kind of concensus among interested parties (result: WhatWG considers SVG outside of their scope). In WML it is not and some implementations accept non-wellformed WML that creates extra problems for others (the motivation is that having two parsers on mobile phone might be a problem due to limited resources, so some prefer to reuse tagsoup parser), as users blame compliant browsers for any parsing errors that occur.

-- 
_______________________________________________
Surf the Web in a faster, safer and easier way:
Download Opera 9 at http://www.opera.com

Powered by Outblaze

Received on Friday, 22 December 2006 10:45:35 UTC