Re: The mapping of phi (was: Re: Exploring new vocabularies for HTML)

That's a nice list of references.   As David said in a separate email thread
about &phi -- you can't win no matter which mapping you choose.

That even sort of applies if you use the Unicode value for the character.
Because Unicode changed the glyph associated with the code, any font that
was developed and not updated before that change has the wrong character at
the &phi and &phiv positions.  So, depending upon the font, the glyph for
3c5 may be a GREEK SMALL LETTER PHI (an "open" curly phi) or GREEK PHI
SYMBOL (a "straight" phi).  See [7] for more details.

Neil Soiffer
Senior Scientist
Design Science, Inc.
www.dessci.com
~ Makers of Equation Editor, MathType, MathPlayer and MathFlow ~



On Sun, Mar 30, 2008 at 2:14 AM, Henri Sivonen <hsivonen@iki.fi> wrote:

> On Mar 30, 2008, at 07:52, Neil Soiffer wrote:
> >       • do you really want to include the >2,000 named entities of
> MathML?
> >       • are their any conflicts with those entities and entities defined
> > in HTML5 -- I'm not aware of any, but has anyone checked?
>
> As it stands, a piece of HTML or XHTML source code with &phi; in it
> changes meaning when it is moved to a document with a MathML public id
> in the doctype or vice versa. Is this the kind of badness that is hit
> with such a low probability that fixing it would do more harm than
> letting it be? (People should be using straight UTF-8 instead of
> entities anyway.) Or is this something that needs fixing? (How?)
>
>  * Mozilla's DTD catalog for XHTML 1.0 and XHTML 1.1 maps &phi; to U
> +03C6 per spec.[1]
>  * HTML5 (text/html) maps &phi; to U+03C6.[2]
>  * Mozilla has updated its DTD catalog for "-//W3C//DTD XHTML 1.1
> plus MathML 2.0//EN", "-//W3C//DTD XHTML 1.1 plus MathML 2.0 plus SVG
> 1.1//EN" and "-//W3C//DTD MathML 2.0//EN" in a way that resolved &phi;
> to U+03D5.[3][4]
>  * The XML Entity definitions for Characters spec maps &phi; to U
> +03D5.[5][6]
>  * The reason for changing what &phi; means is explained in [7] and
> [8].
>  * No one knows if there are JavaScript scripts or CSS selectors out
> there that expect &phi; to expand to U+03C6.
>  * In my personal opinion[14], the list of browser-recognized public
> ids should never again be expanded, in which case XML Entity
> definitions for Characters would not be Web-relevant (except as an
> update to what the MathML 2.0 public ids refer to). [9][10]
>  * Making browsers fetch and parse the actual DTDs is not feasible.
> [11][12]
>  * It seems that this is a real author problem. [13]
>
> [1]
> http://mxr.mozilla.org/seamonkey/source/content/xml/content/src/xhtml11.dtd
> [2] http://www.w3.org/html/wg/html5/#entities
> [3]
> http://mxr.mozilla.org/seamonkey/source/layout/mathml/content/src/mathml.dtd
> [4]
> http://mxr.mozilla.org/seamonkey/source/parser/htmlparser/src/nsExpatDriver.cpp#288
> [5] http://www.w3.org/TR/xml-entity-names/isogrk1.html
> [6] http://www.w3.org/TR/xml-entity-names/isogrk3.html
> [7] http://www.w3.org/Math/characters/
> [8] https://bugzilla.mozilla.org/show_bug.cgi?id=289938#c21
> [9] https://bugzilla.mozilla.org/show_bug.cgi?id=289938#c16
> [10] https://bugzilla.mozilla.org/show_bug.cgi?id=289938#c20
> [11] http://www.w3.org/blog/systeam/2008/02/08/w3c_s_excessive_dtd_traffic
> [12] http://hsivonen.iki.fi/no-dtd/
> [13]
> http://groups.google.com/group/mozilla.dev.tech.mathml/browse_thread/thread/0afaf64eb1b1eab3/d0f26a348c95d4b8
> [14]
> http://groups.google.com/group/mozilla.dev.tech.xml/browse_thread/thread/e7f7efbb5e161348/9fde74f46fb0b5d2
>
> --
> Henri Sivonen
> hsivonen@iki.fi
> http://hsivonen.iki.fi/
>
>
>
> --
> Henri Sivonen
> hsivonen@iki.fi
> http://hsivonen.iki.fi/
>
>
>