[unimath] Fwd: [OS X TeX] STIX Project

[Ross Moore]

Hi all,

This looks promising -- finally!

It looks like they missed making it a birthday present
for me (21 May :-) butif it really is released on Monday,
then that'll be just as good.


Begin forwarded message:

> From: David Derbes <lo...@uchicago.edu>
> Date: 22 May 2010 9:08:29 AM
> To: TeX on Mac OS X Mailing List <macos...@email.esm.psu.edu>
> Subject: [OS X TeX] STIX Project
> Reply-To: TeX on Mac OS X Mailing List <macos...@email.esm.psu.edu>
>
> These folks have been working on a large, free set of technical
> fonts (including LaTeX support) for nearly a decade. They are
> supposed to release version 1 on Monday. Their track record for
> meeting deadlines is not all that good, but we all know software is
> very difficult.
>
> In any case, someone here probably knows a great deal more. This
> might be very good news Real Soon Now.
>
> http://www.stixfonts.org/
>
> David Derbes
> U of Chicago Laboratory Schools


I just glanced through the glyph list.
It looks really nice -- though I guess that I am actually seeing
the beta-version fonts I installed a few years ago -- and they
have really stacked a lot of extra (i.e., non-Unicode) symbols
into the Private Use area.

Have there been any updates to the TeX macro-names recently?
That is, since 2006 for the files:
stix-plane0.tex
stix-plane1.tex

and/or any other tex source files based upon, or extending these.

If so, I really need these now, for my tagged PDF project.
That is, I need to know the (La)TeX macro names that people
can be expected to be using to call upon these characters.


All the best,

Ross

------------------------------------------------------------------------
Ross Moore ross....@mq.edu.au
Mathematics Department office: E7A-419
Macquarie University tel: +61 (0)2 9850 8955
Sydney, Australia 2109 fax: +61 (0)2 9850 8114
------------------------------------------------------------------------

--
You received this message because you are subscribed to the Google Groups "Unicode maths for TeX" group: <http://groups.google.com/group/unimath>. To post to this group, send email to <uni...@googlegroups.com>.

[Will Robertson]

On 22/05/2010, at 11:07 AM, Ross Moore wrote:

> Have there been any updates to the TeX macro-names recently?
> That is, since 2006 for the files:
> stix-plane0.tex
> stix-plane1.tex
>
> and/or any other tex source files based upon, or extending these.

Yes, quite a lot.

> If so, I really need these now, for my tagged PDF project.
> That is, I need to know the (La)TeX macro names that people
> can be expected to be using to call upon these characters.

As you know, I base my list on Barbara Beeton's master version, which I get copies of every now and then. As I understand, it has diverged a little from what I'm using as my master document.

I've been in contact with Aditya Mahajan, who's working on unicode math in ConTeXt, and he's interested in maintaining a shared database from which both platforms can use the same data. His work is at

http://github.com/adityam/context-math

but I can't recall how far along our "shared database" plans ever went.

You and Barbara and I (and Hans, if he's got something to do with unicode-math in ConTeXt; he may not be interested, however) should organise some time to discuss this at TUG2010. It's important we start to use a public master document to base these mappings on. (I.e., a TeX equivalent of David Carlisle's "XML Entity Definitions for Characters".)

The current list of working math characters in the unicode-math package is contained within "unicode-math-table.tex" here:

http://github.com/wspr/unicode-math

As you can see from the script that generates it (stix-extract.sh), there are a few changes, adjustments, and fixes it performs on the current version of the STIX table that am using. (And many characters that are ignored.)

Finally, there are a few ascii symbols that I have inserted control sequences for:

< \UnicodeMathSymbol{"00023}{\octothorpe }{\mathord}{number sign}%
< \UnicodeMathSymbol{"00024}{\mathdollar }{\mathord}{dollar sign}%
< \UnicodeMathSymbol{"00025}{\percent }{\mathord}{percent sign}%
< \UnicodeMathSymbol{"00026}{\ampersand }{\mathord}{ampersand}%
9,12d4
< \UnicodeMathSymbol{"0002C}{\comma }{\mathpunct}{comma}%
< \UnicodeMathSymbol{"0002E}{\period }{\mathpunct}{full stop, period}%
< \UnicodeMathSymbol{"0002F}{\slash }{\mathord}{solidus}%
< \UnicodeMathSymbol{"0003A}{\mathcolon }{\mathpunct}{colon}%
17,18d8
< \UnicodeMathSymbol{"0003F}{\question }{\mathord}{question mark}%
< \UnicodeMathSymbol{"0003F}{\atsign }{\mathord}{commercial at}%

From memory, these are all missing in the STIX table but are somewhat necessary for the unicode-math package. In some cases they're necessary for backwards compatibility as actual input in maths mode (such as \slash and \mathcolon); in other cases the csnames are simply included for completeness (e.g., \question, \atsign).

Will

[Aditya Mahajan]

On Sat, 22 May 2010, Will Robertson wrote:

> On 22/05/2010, at 11:07 AM, Ross Moore wrote:
>
>> If so, I really need these now, for my tagged PDF project.
>> That is, I need to know the (La)TeX macro names that people
>> can be expected to be using to call upon these characters.
>
>
> As you know, I base my list on Barbara Beeton's master version, which I
> get copies of every now and then. As I understand, it has diverged a
> little from what I'm using as my master document.
>
> I've been in contact with Aditya Mahajan, who's working on unicode math
> in ConTeXt, and he's interested in maintaining a shared database from
> which both platforms can use the same data. His work is at
>
> http://github.com/adityam/context-math

That is an old fork of ConTeXt math files and most of those are not
relevant for a shared database.

> but I can't recall how far along our "shared database" plans ever went.

The idea is to have a single file that contains all the data relevant for
unicode math support in all flavors of TeX (cs names, math classes, etc). A
script can then be used to transform this data in a form required by LaTeX and
ConTeXt.

I was using ConTeXt's char-def.lua as a starting point.

http://source.contextgarden.net/tex/context/base/char-def.lua

That file contains a big lua table with the information required by ConTeXt (the
math tables no where near completion).

As a proof of concept, I wrote a lua script which generates
unicode-math-tables.tex that can then be used by Will's unicode math package.
It never moved further than a proof of concept because of time limitation.

> You and Barbara and I (and Hans, if he's got something to do with
> unicode-math in ConTeXt; he may not be interested, however) should
> organise some time to discuss this at TUG2010. It's important we start
> to use a public master document to base these mappings on. (I.e., a TeX
> equivalent of David Carlisle's "XML Entity Definitions for Characters".)

You mean discuss the glyph names? Arriving on a consensus on that will be
difficult. Barabara's list is fairly complete and I think that it should be a
starting point (as opentype-math does).

In ConTeXt, we have been discussing supporting multiple variants of cs names: a
AMS variant (infty, coloneq, etc) and a "full" variant (infinity, colonequals,
etc.)

On a different note, does anyone know if the STIX font is going to be a single
opentype math font or (as it appears in the glyph list) eight or nine opentype
fonts, each covering a partial range?

Aditya

[Ulrik Vieth]

On 05/22/2010 08:56 AM, Aditya Mahajan wrote:
>
> On a different note, does anyone know if the STIX font is going to be
> a single opentype math font or (as it appears in the glyph list)
> eight or nine opentype fonts, each covering a partial range?

If you look at their earlier announcements, the STIX fonts will not be a
single OpenType math fonts, just a collection of glpph containers.

See the entry of January 22, 2010 on http://www.stixfonts.org/news.html:

"As a reminder, this release will be organized in a fashion similar to
the beta test version. After this initial release is made, we will begin
work on a version that incorporates advanced OpenType features so that
version can be compatible for use with Word 2007. A final release will
add LaTeX support. Those releases will follow on as quickly as they can
be developed and tested."

What scares me is when they say "as quickly as they can be developed".
Given their current pace, this could easily mean several more years.

From a practical point of view, the most useful setup of STIX fonts
for LuaTeX might be something like a virtual OpenType math fonts
(similar to the current setup of Latin Modern).

Regards, Ulrik

P.S: Just wondering: Is the virtual OpenType math setup in LuaTeX
specific to Context or is it usable with LuaLaTeX as well?

[Ross Moore]

On 22/05/2010, at 4:56 PM, Aditya Mahajan wrote:

> The idea is to have a single file that contains all the data
> relevant for
> unicode math support in all flavors of TeX (cs names, math classes,
> etc). A
> script can then be used to transform this data in a form required
> by LaTeX and
> ConTeXt.

This sounds like exactly what I am looking for.

What I'm trying to do is automate the interweaving of MathML markup
with the (La)TeX source descriptions of the same piece of mathematics.
Thus for each Unicode entity, I need to know which macro might have
been used to produce it in the TeX version.

I'm doing this with a Perl script, so have an array whose entries
are pieces of regular expressions, which will be used for matching.
e.g.

'21CB' => '\leftrightharpoons',
'21CC' => '\rightleftharpoons',
'21CD' => '(\not\L|\nL)eftarrow',
'21CE' => '(\not\L|\nL)eftrightarrow',
'21CF' => '(\not\R|\nR)ightarrow',
'21D0' => '\Leftarrow',
'21D1' => '\Uparrow',
...
'1D466' => '(\it)y?',
'1D467' => '(\it)z?',
'1D468' => '\b(fit|m|oldsymbol)(A|{A})',
'1D469' => '\b(fit|m|oldsymbol)(B|{B})',
'1D46A' => '\b(fit|m|oldsymbol)(C|{C})',
...
'1D536' => '\(math)?frak(A|{y})',
'1D537' => '\(math)?frak(A|{z})',
'1D538' => '\(math|B)bb(A|{A})',
'1D539' => '\(math|B)bb(B|{B})',
'1D53A' => '\(math|B)bb(C|{C})', # non-unicode, see '2102'
'1D53B' => '\(math|B)bb(D|{D})',

>
> I was using ConTeXt's char-def.lua as a starting point.
>
> http://source.contextgarden.net/tex/context/base/char-def.lua
>
> That file contains a big lua table with the information required by
> ConTeXt (the
> math tables no where near completion).
>
> As a proof of concept, I wrote a lua script which generates
> unicode-math-tables.tex that can then be used by Will's unicode
> math package.
> It never moved further than a proof of concept because of time
> limitation.

OK. So I'll need to get a copy of this unicode-math-tables.tex
and merge what it has with my own array.

>
>> You and Barbara and I (and Hans, if he's got something to do with
>> unicode-math in ConTeXt; he may not be interested, however) should
>> organise some time to discuss this at TUG2010. It's important we
>> start to use a public master document to base these mappings on.
>> (I.e., a TeX equivalent of David Carlisle's "XML Entity
>> Definitions for Characters".)
>
> You mean discuss the glyph names? Arriving on a consensus on that
> will be
> difficult.

I don't see any need to agree on single names for each symbol.
What is important. however, is to avoid name clashes.

The problem that I see looming with the release of the STIX fonts,
is what are the MathML/XML guys going to do to support all those
(~1000) new characters in the Private Use area?
I suppose it just won't be so Private anymore.

> Barabara's list is fairly complete and I think that it should be a
> starting point (as opentype-math does).

Certainly. As this figures to be what the bigger math publishers
will most likely be using, for the STIX fonts.
Also, it figures to be reasonably back-compatible with existing names
used with earlier (Type 1) fonts and their supporting packages.

>
> In ConTeXt, we have been discussing supporting multiple variants of
> cs names: a
> AMS variant (infty, coloneq, etc) and a "full" variant (infinity,
> colonequals,
> etc.)

Yes. This is what I need too.
Ultimately I want to be able to process existing LaTeX source
and produce fully tagged PDF, without the need to change anything
in the document source at all. It will just be a different TeX
engine and an enriched format file, doing the extra work.

>
> On a different note, does anyone know if the STIX font is going to
> be a single
> opentype math font or (as it appears in the glyph list) eight or
> nine opentype
> fonts, each covering a partial range?

I've seen nothing to suggest anything other than multiple fonts.


>
> Aditya

All the best,

Ross


------------------------------------------------------------------------
Ross Moore ross....@mq.edu.au
Mathematics Department office: E7A-419
Macquarie University tel: +61 (0)2 9850 8955
Sydney, Australia 2109 fax: +61 (0)2 9850 8114
------------------------------------------------------------------------

[Aditya Mahajan]

On Sat, 22 May 2010, Ulrik Vieth wrote:

> On 05/22/2010 08:56 AM, Aditya Mahajan wrote:
>>
>> On a different note, does anyone know if the STIX font is going to be
>> a single opentype math font or (as it appears in the glyph list)
>> eight or nine opentype fonts, each covering a partial range?
>
> If you look at their earlier announcements, the STIX fonts will not be a
> single OpenType math fonts, just a collection of glpph containers.
>
> See the entry of January 22, 2010 on http://www.stixfonts.org/news.html:
>
> "As a reminder, this release will be organized in a fashion similar to the
> beta test version. After this initial release is made, we will begin work on
> a version that incorporates advanced OpenType features so that version can be
> compatible for use with Word 2007. A final release will add LaTeX support.
> Those releases will follow on as quickly as they can be developed and
> tested."

Thanks.

> From a practical point of view, the most useful setup of STIX fonts
> for LuaTeX might be something like a virtual OpenType math fonts
> (similar to the current setup of Latin Modern).

No. The most useful setup is to use the xits fonts by Khaled :-).

> P.S: Just wondering: Is the virtual OpenType math setup in LuaTeX specific to
> Context or is it usable with LuaLaTeX as well?

Do you really need a virtual OpenType font setup with LuaLaTeX? In
ConTeXt, we need virtual fonts because MkIV has no support for type1
fonts. LuaLatex is based on LaTeX2e format, so type1 fonts can be used out
of the box. With unicode-math package, opentype math fonts can be used. So
what is the need for a virtual math font setup, STIX fonts? I do not think
that ConTeXt will provide a virtual math font for STIX, simply because
xits exists and I am sure Khaled will update xits when the new STIX fonts
are released.

The virtual fonts will not work out of the box in LuaLaTeX, but it should
be simple to support them. Most of the data (char-def.lua, math-vfu.lua,
math-map) can be reused; math-ini.lua, which actually defines the
mappings, uses luatex primitives (\unexpanded\xdef...{\Umathchar...},
etc), so it should also work with LaTeX. The only thing that needs to be
done is to hook the font callbacks in the appropriate place. ConTeXt does
this in font-def.lua, and hooks this to the typescript mechanism. I do not
understand LaTeX font handling well enough to know where this should be
done in LaTeX.

Aditya

[Aditya Mahajan]

On Sun, 23 May 2010, Khaled Hosny wrote:

> On Sat, May 22, 2010 at 10:50:27PM +0200, Ulrik Vieth wrote:
>> On 05/22/2010 08:56 AM, Aditya Mahajan wrote:
>>>
>>> On a different note, does anyone know if the STIX font is going to be
>>> a single opentype math font or (as it appears in the glyph list)
>>> eight or nine opentype fonts, each covering a partial range?
>>
>> If you look at their earlier announcements, the STIX fonts will not
>> be a single OpenType math fonts, just a collection of glpph
>> containers.
>>
>> See the entry of January 22, 2010 on http://www.stixfonts.org/news.html:
>>
>> "As a reminder, this release will be organized in a fashion similar
>> to the beta test version. After this initial release is made, we
>> will begin work on a version that incorporates advanced OpenType
>> features so that version can be compatible for use with Word 2007. A
>> final release will add LaTeX support. Those releases will follow on
>> as quickly as they can be developed and tested."
>>
>> What scares me is when they say "as quickly as they can be developed".
>> Given their current pace, this could easily mean several more years.
>

> Creating an OpenType MATH version based on the released beta took me
> like 2 hours to get a basic working version, so lets not worry about
> that :)

I am hoping that you will spend 2 more hours on that :-)

Aditya

[Khaled Hosny]

On Sat, May 22, 2010 at 10:50:27PM +0200, Ulrik Vieth wrote:

> On 05/22/2010 08:56 AM, Aditya Mahajan wrote:
> >
> >On a different note, does anyone know if the STIX font is going to be
> > a single opentype math font or (as it appears in the glyph list)
> > eight or nine opentype fonts, each covering a partial range?
>
> If you look at their earlier announcements, the STIX fonts will not
> be a single OpenType math fonts, just a collection of glpph
> containers.
>
> See the entry of January 22, 2010 on http://www.stixfonts.org/news.html:
>
> "As a reminder, this release will be organized in a fashion similar
> to the beta test version. After this initial release is made, we
> will begin work on a version that incorporates advanced OpenType
> features so that version can be compatible for use with Word 2007. A
> final release will add LaTeX support. Those releases will follow on
> as quickly as they can be developed and tested."
>
> What scares me is when they say "as quickly as they can be developed".
> Given their current pace, this could easily mean several more years.

Creating an OpenType MATH version based on the released beta took me
like 2 hours to get a basic working version, so lets not worry about
that :)

> From a practical point of view, the most useful setup of STIX fonts
> for LuaTeX might be something like a virtual OpenType math fonts
> (similar to the current setup of Latin Modern).
>
> Regards, Ulrik
>
> P.S: Just wondering: Is the virtual OpenType math setup in LuaTeX
> specific to Context or is it usable with LuaLaTeX as well?

ConTeXt only, for now at least.

Regards,
Khaled

--
Khaled Hosny
Arabic localiser and member of Arabeyes.org team
Free font developer

[Taco Hoekwater]

Aditya Mahajan wrote:
>
>> From a practical point of view, the most useful setup of STIX fonts
>> for LuaTeX might be something like a virtual OpenType math fonts
>> (similar to the current setup of Latin Modern).
>
> No. The most useful setup is to use the xits fonts by Khaled :-).

Is this assuming the actual release will look like the beta?

>> P.S: Just wondering: Is the virtual OpenType math setup in LuaTeX
>> specific to Context or is it usable with LuaLaTeX as well?

As Aditya and Khaled wrote: not immediately. Hans and I are thinking
of a way to dump the generated virtual font to a lua table as this
was (also) requested by the polish font foundry, and that may make
it usable by other luatex formats.

Best wishes,
Taco

[Khaled Hosny]

On Sun, May 23, 2010 at 07:43:29AM +0200, Taco Hoekwater wrote:
> Aditya Mahajan wrote:
> >
> >>From a practical point of view, the most useful setup of STIX fonts
> >>for LuaTeX might be something like a virtual OpenType math fonts
> >>(similar to the current setup of Latin Modern).
> >
> >No. The most useful setup is to use the xits fonts by Khaled :-).
>
> Is this assuming the actual release will look like the beta?

It doesn't really matter as long as they keep assigning proper Unicode
code points to the glyphs.

Regards,
Khaled

--
Khaled Hosny
Arabic localiser and member of Arabeyes.org team
Free font developer

[Khaled Hosny]

On Sat, May 22, 2010 at 09:37:57PM -0400, Aditya Mahajan wrote:
> On Sun, 23 May 2010, Khaled Hosny wrote:
>
> >Creating an OpenType MATH version based on the released beta took me
> >like 2 hours to get a basic working version, so lets not worry about
> >that :)
>
> I am hoping that you will spend 2 more hours on that :-)

You have my promise of a working beta within a week of the release, as
long as I'm still alive and able to use computers by then ;)

Regards,
Khaled

--
Khaled Hosny
Arabic localiser and member of Arabeyes.org team
Free font developer