[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Stix/Unicode/MathML/mathfont glyph tables



Taco Hoekwater wrote:

> - Finally: I'd be very, very pleased if somebody would send me a zip 
>   file of the HTML+gif tables from STIX and MathML. I can only fetch 
>   them from the WWW the hard way using a regular browser, and "save 
>   image as" for a couple thousand images does not sound funny at all.

As for MathML getting the whole tree of HTML and GIFs should be easy:

  http://www.w3.org/TR/REC-MathML-19980407.zip
  http://www.w3.org/TR/REC-MathML-19980407.tar.gz

As for the STIX tables, I'd recommend GNU wget, which is a handy
utility for downloading or mirroring whole trees of WWW pages. 

Basically, what you need is:

  wget --mirror -A.html -A.pdf -A.gif -o mirror.log
       --http-user=STIX --http-passwd=Char0n http://www.ams.org/STIX/

However, this may take a long time (several hours) due to the vast
number of tiny files that have to be retrieved.

If you just want to get the HTML pages, you could also try this:

  wget --mirror -A.html -o mirror.log
       --http-user=STIX --http-passwd=Char0n http://www.ams.org/STIX/
         

Now for the rest of your message:

> The problem is in the first section. There are just to many tables 
> and I am quite uncertain which ones I need to merge to get the 
> desired overall table. The 35 Stix submission tables obviously do not 
> contain all glyphs (I assume the ones that are left out are the ones 
> that are already in Unicode). 

I suppose so, but I'm not sure of that myself.

> I really need an editable table of all chars so as to generate an 
> overall index (so that I can later insert font/glyph codes). This 
> means I need to sort out all these existing tables, and that leads 
> to some questions:

> - Is ISO 9573-13 a full subset of STIX (guess so)?

> - What about the things in 8859's Annex that are not in 9573-13?

Is there any good reference as to what exactly is in ISO 9573-13?
And what's this 8859 Annex anyway?  

> - Am I correct in my assumption that MMEXTRA no longer exists
>   (or only defines non-marking entities)?

As of REC-MathML-19980407, the MMEXTRA tables seem to empty indeed.

> - Do the Unicode Reference Blocks (from the STIX site) contain *all* 
>   math-related glyphs from current Unicode?

I suppose so.

> - How to read those blocks? What are the squares, and why are some   
>   of the squares empty?

Obviously some of the glyph images seem to be missing on the AMS
server.  Same problem for one of the PDF files (stixiso-21.pdf).

> - If I merge the Reference Blocks with the STIX private blocks, does 
>   that give me *all* glyphs?

I suppose so.  However, this list of Unicode/STIX glyphs doesn't
necessarily have a one-to-one mapping to font characters needed for
TeX.  Just consider the big operators, delimiters and wide accents
which are represented by single control sequence, but are constructed
from multiple font glyphs.

> - Are the private indices from MathML the same as for STIX?

I would guess so given that MathML makes reference to the STIX project
and was put together by the same people (i.e. Patrick and Barbara).
However, it certainly doesn't hurt to double-check this assumption.

> - I'm not sure, but I seem to remember from a previous investigation 
>   that there were glyphs in mathfont that do not appear in the STIX 
>   tables?

I don't know for sure, but this is certainly possible since the
mathfont tables were developed largely indepently of the STIX tables.  
In case you want to make a list of these glyphs, however, you'd have
to distinguish between two kinds of glyphs:

- glyphs appearing in mathfont that represent new kinds of symbols.
  These glpyhs presumably should be added to STIX, provided that
  there are valid references for their use.

- glyphs appearing in mathfont only for technical reasons, e.g.
  building blocks for big delimiters or big variants of operators.
  These glyphs presumably don't belong into a glyph standard such
  as Unicode or STIX since they don't represent new symbols by
  themselves, but they are needed in a font standard for TeX.

It is obvious that this discrepancy between glyphs font characters
will probably lead to some problems during the implementation of a
reference font set.  Without having spent much time thinking about it,
I am not really sure myself how to proceed.  Maybe it would be best 
to have a set of "glyph containers" organized by Unicode/STIX code
postitions and a few "private" pages of TeX-specific glyphs, which
might then be used to building blocks for usable TeX fonts.

Just a few comments.

Cheers, Ulrik.

P.S.  I'm still very busy finishing my thesis, so I really can't
afford to spend much time on math fonts right now.