Re: [Texmacs-dev] epsilon in html

[David Allouche]

On Fri, 2005-07-22 at 14:56 +0200, Joris van der Hoeven wrote: 
> On Fri, Jul 22, 2005 at 12:08:09PM +0200, Bas Spitters wrote:
> > But I now understand where confusion comes from.
> > What is called \varepsilon in latex is the ordinary ε in html.
> > What is called \epsilon in latex is the lunate epsilon (03F5) in html.

Right... that looks like a bug.

I compiled most of the data by correlating various sources, like texmacs
symbol names, MathML entity definitions and Omega conversion tables.

In Omega you can find
./cmmi.onm:19:\SGMLFontEntity{cmmi}{"0F}{\SGMLname{epsilon}}{mi}{}%
./cmmi.onm:38:\SGMLFontEntity{cmmi}{"22}{\SGMLname{varepsilon}}{mi}{}%

The corresponding TeX characters are mapped to symbols in texmacs
./cmmi.enc:1:11  "alpha" "beta" "gamma" "delta" "epsilon"
./cmmi.enc:5:    "psi" "omega" "varepsilon" "vartheta"

When you collate those tables, you find that according to Omega, the
TeXmacs <epsilon> symbol should be MathML &epsilon; and <varepsilon>
should be &varepsilon;

In MathML you find:

./xhtml-math11-f.dtd:9728:<!ENTITY varepsilon       "&#x003B5;" ><!--alias 
ISOGRK3 epsiv -->
./html/symbol.ent:37:<!ENTITY epsilon          "&#x003B5;" ><!-- -->

In other words, both <epsilon> and <varepsilon> are mapped to U+003B5 by
collating TeXmacs, Omega, and MathML tables. So external data is of
little help to figure out the right thing, since clearly at least one
link of the chain is wrong.

It seems to me that the <epsilon> and <varepsilon> texmacs symbols
should be swapped: the proper greek letter epsilon should be <epsilon>,
and the alternate fancy symbould should be <varepsilon>.

In my opinion that's not a bug in the translation tables, but a bug in
the texmacs symbol naming. Well, now that makes two bugs, symbol naming
_and_ conversion table. I guess you could call that a fix.

-- 
                                                            -- ddaa

signature.asc
Description: This is a digitally signed message part