Mozilla 1.9 MathML status and notes
Karl Tomlinson
(http://www.stixfonts.org/), which provide the best support for
mathematical characters.
If users have previously set the "font.mathfont-family"
preference, then it is best to reset this to the default value.
(URL about:config -> context menu -> Reset)
A Beta release of the STIX fonts is available under this license:
http://www.mozilla.org/projects/mathml/fonts/stix/beta-license.txt
and may be downloaded from
http://www.mozilla.org/projects/mathml/fonts/stix/STIXBeta.zip
Positioning is currently (very) incorrect with MathML and tables.
This includes <mtable> which is used for matrices, conditionals,
and sometimes alignment of systems of equations. There is a patch
in this bug
https://bugzilla.mozilla.org/show_bug.cgi?id=363240
that resolves this issue, and Bill Gianopoulos has provided some
unofficial Windows and Linux builds with this patch here:
http://www.wg9s.com/mozilla/firefox/
On MS Windows, sometimes hex boxes are drawn instead of the glyphs
from the STIX fonts:
https://bugzilla.mozilla.org/show_bug.cgi?id=413115
https://bugzilla.mozilla.org/show_bug.cgi?id=382542
Please file a bug if you notice any issues not mentioned here.
BaKoMa TrueType Computer Modern fonts are now normally not used
because their encoding is inconsistent with Unicode
recommendations.
https://bugzilla.mozilla.org/show_bug.cgi?id=161137#c39
https://bugzilla.mozilla.org/show_bug.cgi?id=400938#c0
However, if a document explicitly requests font-family CMSY10 or
CMEX10 or similar, and the BaKoMa fonts are installed, then these
fonts get used, which will likely result in the wrong symbols
being drawn. Probably the best work-around for this is to
uninstall these fonts.
If the Computer Modern font family is preferred, then the best
version of this font that I know to recommend is the PostScript
Type 1 implementation by Blue Sky Research and Y&Y, Inc.
http://www.ams.org/tex/type1-fonts.html
I haven't tested these. They may work reasonably well on Unix,
but I'm not sure about other platforms.
William F Hammond
> BaKoMa TrueType Computer Modern fonts are now normally not used
> because their encoding is inconsistent with Unicode
> recommendations.
>
> https://bugzilla.mozilla.org/show_bug.cgi?id=161137#c39
> https://bugzilla.mozilla.org/show_bug.cgi?id=400938#c0
>
> However, if a document explicitly requests font-family CMSY10 or
> CMEX10 or similar, and the BaKoMa fonts are installed, then these
> fonts get used, which will likely result in the wrong symbols
> being drawn. Probably the best work-around for this is to
> uninstall these fonts.
1. Clarification: Among all the font collections previously recommended
at projects/mathml/fonts you are recommending uninstalling only those
in the "deceitful font" category, as described in Bugzilla/400938.
That is, the other (categories 2 and 3) non-unicode fonts will be OK.
Is this correct?
2. I trust that font casting via "mathvariant" will remain supported.
( For example, '<mi mathvariant="fraktur">A</mi>' )
Don't forget legacy MathML, v.2, content.
Thanks.
-- Bill
Karl Tomlinson
> 1. Clarification: Among all the font collections previously recommended
> at projects/mathml/fonts you are recommending uninstalling only those
> in the "deceitful font" category, as described in Bugzilla/400938.
> That is, the other (categories 2 and 3) non-unicode fonts will be OK.
> Is this correct?
I was recommending uninstalling fonts in the "Deceitful Fonts"
category. They are probably only going to cause problems if a
document explicitly requests these fonts, so it may not be
_necessary_ to uninstall them.
Category 2 "PUA-only Unicode fonts" won't cause a problem.
Category 3 "Fonts with a Microsoft Symbol charmap" may cause
similar problems. Different platforms treat these fonts
differently, and there is disagreement about what Unicode points
the glyphs in these characters should be mapped to.
https://bugzilla.mozilla.org/show_bug.cgi?id=399636
These fonts have a Macintosh Roman cmap as well as the "Unicode,
non-standard" cmap, and Linux (and maybe Mac) use this cmap.
The problem is the glyphs are not representative of the code
points in the Roman cmap.
Even on MS Windows which uses the "Unicode, non-standard" cmap,
many applications (or maybe its the OS libraries) map the PUA
Unicode points to the Latin Unicode block (which results in a
mapping slightly different to that obtained by mapping the
Macintosh Roman charset to Unicode).
http://www.microsoft.com/typography/otspec/recom.htm#sym
At least with these Microsoft Symbol fonts, we could detect them,
and choose a behavior. I have my thoughts on what the appropriate
Unicode mappings should be, but the problem is that there is more
than one opinion (or expectation) on this.
I don't expect or suggest that MS Windows should uninstall their
SymbolMT font, so my best advice is the web authors should not
request the "Symbol" font family.
Karl Tomlinson
> 2. I trust that font casting via "mathvariant" will remain supported.
> ( For example, '<mi mathvariant="fraktur">A</mi>' )
> Don't forget legacy MathML, v.2, content.
Unfortunately this hasn't yet been implemented in Mozilla.
(I don't think this ever was supported in Mozilla.)
https://bugzilla.mozilla.org/show_bug.cgi?id=114365
Some web pages may do their best to make things render in a
fraktur style, by using CSS rules to set font families based on
the attribute.
William F Hammond
Many thanks for all of the answers I now find from you.
> I don't expect or suggest that MS Windows should uninstall their
> SymbolMT font, so my best advice is the web authors should not
> request the "Symbol" font family.
I've not seen "Symbol" much requested in web pages. Rather, as I recall,
"Symbol" has been one of the names pushed in "res/mathml.css" and,
as appropriate, featured in the missing fonts requester that appears
the first time during a browser invocation that a page with MathML
is loaded.
-- Bill
William F Hammond
>> 2. I trust that font casting via "mathvariant" will remain supported.
>> ( For example, '<mi mathvariant="fraktur">A</mi>' )
>> Don't forget legacy MathML, v.2, content.
>
> Unfortunately this hasn't yet been implemented in Mozilla.
> (I don't think this ever was supported in Mozilla.)
Yes, I think you are correct. Sorry about my lapse.
I've been running with something in userContent.css like:
[mathvariant="script"] {
font-family: "Euclid Math One", serif;
}
[mathvariant="fraktur"] {
font-family: "Euclid Fraktur", serif;
}
[mathvariant="double-struck"] {
font-family: "Euclid Math Two", serif;
}
Modulo font availability and distributability I suppose this kind
of arrangement could be the default.
The W3 Math WG still has not said much about authoring. But, for
example, let me suggest that most of what's out there comes, one way
or another from the world of LaTeX, and in that world I believe
font casting is the most common way of marking up for fraktur,
calligraphic, and double-struck. The sensible thing in this case for
a program like tex4ht (when called for XHTML+MathML generation) is
to use the "mathvariant" technique for font casting.
-- Bill