aliftype / xits Goto Github PK

Hi Khaled,

I had a closer look at the OT math parameters and have one question:
Why is RadicalRuleThickness (66) bigger than Overbar/UnderbarRuleThickness (50)?
Changing RadicalRuleThickness would also make it consistent with RadicalExtraAscender.

I would have expected that all rule thicknesses were suppsed to be consistent.

Regards, Ulrik

Extensible glyphs show small gaps between its parts in MS Office, Cambria does not suffer from this so there might be a way to fix it.

I'm using XITSMath font with lualatex and unicode-math.

Sample code:

\documentclass{article}
\usepackage{amsmath}
\usepackage{unicode-math}
\setmathfont{XITSMath}
\begin{document}
$\binom{n}{m}$
\end{document}

Then spacing between n and m looks too large; m is positioned at outside of parentheses.

For compatibility with unicode-math & XeTeX, adding a second scriptstyle (ssty) feature for the prime would be helpful... See latex3/unicode-math#210

These features will making any digits into subscript and superscript forms if users want to do and software supports such operatings. STIX 2.0 have already implemented them.

XITS Math 7.1 is missing some of the subscript letters. It supports a,e,u,o,r,u,x. Missing are h,j,k,l,m,n,o,s,t.

Example tex file:

\documentclass{article}
\usepackage{agda}
\setmathfont{xits-math.otf}


\begin{document}
a $xₐ$\\
e $xₑ$\\
h $xₕ$\\
i $xᵢ$\\
j $xⱼ$\\
k $xₖ$\\
l $xₗ$\\
m $xₘ$\\
n $xₙ$\\
o $xₒ$\\
p $xₚ$\\
r $xᵣ$\\
s $xₛ$\\
t $xₜ$\\
u $xᵤ$\\
x $xₓ$\\
\end{document}

Resulting pdf when rendered with xelatex: http://files.314.ch/Test.pdf

See also https://en.wikipedia.org/wiki/Superscripts_and_Subscripts and http://www.unicode.org/charts/PDF/U2070.pdf

See https://en.wikipedia.org/wiki/Mathematical_alphanumeric_symbols

STIX bar glyph (|) is drawn almost on base line, it looks a bit ugly in equations like $|\{x\in\mathbb{Q}\mid x^2<2\}|$, (see this typophile discussion). The preference for bar varies from font to another (see this typophile discussion, too). Checking some Times-like font, I found both usages, but since STIX bar looks ugly in math (and is not consistent with the double bar), I think we should change it in math font at least.

From the 'rtlm' tag definition:

"This feature applies mirrored forms appropriate for right-to-left text other than for those characters that would be covered by the character-level mirroring step performed by an OpenType layout engine. "

Hi,

I am using the XITS 1.106 with MS Word v14.0.6123 on window 7.

When I print a document with XITS to PDF from word, the pdf that is created does not show the XITS fonts properly. Large parts of the text are missing.

How can I get the times maths fonts to print properly?

Thanks,
Charlie

Are there plans to release Webfont kit? That could include EOT, SVG and WOFF formats. And/or publish XITS and XITS Math on Google Fonts.

SIL OFL License allows conversion in WOFF under some complex conditions, but they are difficult to comply. And other formats can't be reached without change of font name.

As far as I understand there is currently no italic bold math font. Is that something that could be developed? The use case is some matrix notation (ISO/IEC convention). Or is there a workaround?

Extensible glyphs show small gaps between its parts in MS Office, Cambria does not suffer from this so there might be a way to fix it.

We need a stylistic set (or character variant?) to map from the default form to the other.

We need a user manual that covers installation for different OSs/engines/macro packages, different font options and some usage example.

Any help writing such manual is highly appreciated.

Some packagers (e.g. fedora) would appreciate a standard make file, we would also need it to support more build options.

I'm using XITS with Word 2013. If a document contains product signs (from the Large Operator menu in Equation tools), the product signs are mangled when the document is converted to PDF. For example, these equations in Word:

are rendered like this in the PDF:

(The first equation is a numbered equation in a table, while the second is a bare equation on its own line. The equations are in 12-point font, though changing the font size doesn't seem to affect the results.)

This doesn't happen if I use Cambria Math, so it appears to be a XITS issue, not a Word issue. Fortunately, summation signs seem to work fine.

This feature would making these chatacters useful to produce certain case as x²ⁿ₀, see the behavior in Libertinus Serif.

If I use the Word 2013 export as PDF feature, some symbols are missing (see attachments). If I create a PDF using the Microsoft print to PDF function, everything looks fine.
bug-in-xits.pdf
bug-in-xits-print.pdf

This is obviously a missing feature in STIX, but just for the record and in case we want to add old style numbers in XITS before.

Some fonts like Neo Euler and Asana Math provide these glyphs via the "onum" feature:
http://fred-wang.github.io/MathFonts/

Script glyphs from STIX are spaced as if they were upright but they are in fact slanted. This also makes *scripts positioned in odd ways. I usually avoid changing STIX glyph matrices, but in this case there is no other work around.

Khaled,

Given that STIX 2 has a math font, what's the status of this project. Can you please clarify?

Double-struck bold upright variants for the following letters: C, D, N, P, Q, R, Z, d, e, i, j are missing in XITS Math, so stylistic set ss05 don't provide complete alphabet. I temporarily use they from XITS Math Bold, but I hope they could be in the main font.

We should provide more sizes of the hat and tilde.

When inserting a vertical line as right delimiter of a fraction in Word 2013, the subscript of xits-math (xits-1.106) is set too high (see attachment).
bug-in-xits.pdf

Check the PDF's at http://folk.ntnu.no/mitrevsk/temp/xits_bug/

The problem appears with both XeLaTeX and LuaLaTeX (using TL2010 pretest in Windows7).

I guess because we have very high ascent/descent otherwise large glyphs are cropped.

Hi, Khaled.

Your work here is impressive and I see that you have distilled a lot of knowledge with the creation of the XITS fonts.

As a by-product of this, could you create a document (well, anything at all is better than no document) on how one would go adding support for mathematics to a font? Ideally, this could be converted, as much as possible, in a set of scripts, with the fine-tuning being left for humans?

I would be one of the first persons to apply your expertise to some other fonts as an experiment.

Thanks so much for the work you've been doing.

FireFox does not support OpenType math, yet, it instead relays on per font configuration files. We can generate such files by a python script that reads the exiting math data and output Mozilla configuration file. However, FireFox's implementation takes code points not glyph names or even indices, so we will need to assign unencoded glyphs PUA code points first.

Please see this luatex bug report where I provide examples of how lualatex produces very poor limits spacing when typesetting integrals in display style.

Hans suggests that it is a lack of appropriate OpenType kerning information; I don't understand the details here, but I have noticed that XeLaTeX is able to typeset the integral without difficulty.

I don't know whether this is easy to fix or not.

Thanks!

Stylistic sets don't work in Microsoft Word. I've tried double-struck and calligraphic letters, and no stylistic set allows me to use their different variants.

I use the following guide: http://office.microsoft.com/en-us/word-help/opentype-options-in-the-font-dialog-box-HA101809106.aspx
I've tested font Gabriola, and stylistic sets work for it, so maybe this is a problem of XITS-Math.

Microsoft Word 2013.

While there are some rudimentary and effective guides for enabling XITS math support for ConTeXt and LaTeX (based on XeTeX or LuaTeX engines), there is no good guide that I can find which demonstrates how to correctly and completely switch to using XITS for both math and regular rendering in a plain TeX document that uses either XeTeX or LuaTeX. This would be most welcome.

i use word 2010 an telive 2010 but xits font not good for vietnamese.
ex: ố, ự, .......

I'm just opening this follow-up of #36 for the record. See #36 (comment)

• redesign the glyphs
• Add U+1EEF0 and U+1EEF1
• Make U+1EEF0 and U+1EEF1 stretchable

Some letters (namely, from U+21xx range) aren't substituted when ss06 or ss07 variants are selected. Proposed patch changes this behavior: http://pastebin.com/5WjBWGgT
I've tested it with XeLaTeX and unicode-math.

ConTeXt minimal example:

\switchtobodyfont [xits]
\starttext
compare $1/2$ with $1\,/\,2$ etc.
\stoptext

Seems to be a font issue, not a particular ConTeXt issue.

Thanks

Like in the pre STIXv1.0.0 version, we should add upright integrals as a stylistic set.

Whis this feature the width of digits would be proportional (especially 1), this feature can also be added over 'onum' feature.

Setting bold-style=ISO in unicode-math, and an example of $R(\mathbf{p})$, it is easily seen that the thickness of bold italic p is almost the same as non-bold R. When there is no non-bold p arround, it can hardly be inferred that \mathbf{p} here refers to a vector rather than a scalar.

The current over/under braces are inconsistent between pre-built sizes and the composite ones, we need a new consistent set.

XITS Math lacks kerning, which means the spacing between letters do not seem uniform. A simple $ABCDEFGabcdefg$ with unicode-math will reveal the problem.

Currently one can use this temporary fix: instead of a single line

\setmathfont{XITS Math}

we can use three lines

\setmathfont{XITS Math}
\setmathfont[range={\mathit}]{XITS Italic}
\setmathfont[range={\mathrm}]{XITS}

Is there a way to make XITS fonts under Windows using make?
I see Makefile, and I have the latest sortsmill Python package. But my sortsmill has no module ffcompat. So, make is unsuccessful.
I also have FontForge, and there is no significant Python code too.
Where could I get ffcompat?

Typically, the xits-specimen file should contain more textual material, text and display math, chemical formulas and other scientific materials.

I try the following:

\usetypescript[xits]
\setupbodyfont[xits]
\starttext
$σ$
\stoptext

No σ in the resulting PDF file.

\underbrace and likes are too close to the base line.

I'm using XITS Version 1.012 (according to FONTLOG.txt), and XeLaTeX via LyX and MiKTeX on Windows.

In my opinion, there is rather too much spacing around integral symbols. This is particularly apparent around double (or more) integrals, because they seem to be very tight together (even by the standards of double integral symbols). For instance, the following looks rather wrong to me:

\left(\frac{a}{2}\right)^2 \iint f(x,y) \,\mathrm{d}x \,\mathrm{d}y

I'm using XITS Version 1.012 (according to FONTLOG.txt), and XeLaTeX via LyX and MiKTeX on Windows.

If I put a tilde on a Latin capital A, like normal, everything works fine:

\tilde{A} = \dotsb

The tilde is fairly small, and is centered on the top point of the A even though this is not the center of the overall character. (By the way, this is not true of Latin capital I: there the tilde is incorrectly centered on the character's bounding box.)

Things go haywire if I add a superscript, even though it's outside the tilde command:

\tilde{A}_\alpha = \dotsb

Now the tilde is much wider, and centered on A's bounding box rather than its top point. I have no idea how the tilde command could even "know" that there is a superscript outside. Perhaps this is a bug in unicode-math, since XITS is the only unicode font I've tried. However it doesn't occur with XeTeX and Computer Modern, nor with any LaTeX font I've tried. It's possible to work around it by using an mbox:

\mbox{\ensuremath{\tilde{A}}}_\alpha = \dotsb

Obviously this isn't ideal. Strangely the subscript kerns correctly, even though I would not expect it to know what's happening inside the mbox.

The radical bar is separated from the radical glyph.

Is it possible to add N'ko language support in xits-math ?
In n'ko, we use the mirrored version of math symbols because N'ko is a RTL alphabet like arabic.
Here are some links :
en.wikipedia.org/wiki/N%27Ko_alphabet
http://www.kanjamadi.com/
http://www.unicode.org/charts/PDF/U07C0.pdf

We should consider a bold math font that can be used to get bold symbols.