retex/rex

ReX - typesetting mathematics

rex's Issues

Convert `kerning_table` and `variant_tables` to an array

We would be better of using a binary search for these two tables instead of a hashmap, since they are so small.

Notice the compile size of each module

rw-r--r-- 0/0   8476 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-dimensions.o
rw-r--r-- 0/0  39136 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-font.o
rw-r--r-- 0/0   5328 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-font-constants.o
rw-r--r-- 0/0   8420 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-font-fontselection.o
rw-r--r-- 0/0 137592 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-font-glyphs.o
rw-r--r-- 0/0  13312 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-font-kerning.o
rw-r--r-- 0/0 126672 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-font-kerning_table.o
rw-r--r-- 0/0   1468 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-font-offsets.o
rw-r--r-- 0/0  83956 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-font-symbols.o
rw-r--r-- 0/0 308764 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-font-variant_tables.o
rw-r--r-- 0/0  26452 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-font-variants.o
rw-r--r-- 0/0  65312 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-functions.o
rw-r--r-- 0/0  34940 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-layout.o
rw-r--r-- 0/0  11352 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-layout-builders.o
rw-r--r-- 0/0  96844 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-layout-engine.o
rw-r--r-- 0/0   3260 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-layout-spacing.o
rw-r--r-- 0/0  65672 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-lexer.o
rw-r--r-- 0/0   1100 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-lexer-lexer.o
rw-r--r-- 0/0   3452 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-parser.o
rw-r--r-- 0/0   4428 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-parser-atoms.o
rw-r--r-- 0/0   9056 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-parser-builders.o
rw-r--r-- 0/0  86720 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-parser-engine.o
rw-r--r-- 0/0  41780 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-parser-nodes.o
rw-r--r-- 0/0  47668 Jan  1 01:00 1970 rex-43e78ccd3d61ee32.rex-render.o

Migrate HashMaps to PHF build script

In particular, migrate symbols.rs and glyphs.rs to codegen from a build.rs script. See this example: https://github.com/nagisa/target_build_utils.rs/blob/master/build.rs#L26. We can place the data required to generate these in a json file.

#[no_std]

Make it possible to compile without std.

• introduce feature use_std

Motivation:

• remove file system dependency
• remove stdin/stdout/stderr
• smaller binary

Need a testing framework

There is some prior work done here for related projects:

• KaTeX uses a selenium docker to render their content in firefox and chrome.
• KaTeX also can compare LaTeX with their output for comparisons.
• MathJax also uses selenium, with a Web UI it appears
• WebKit has documents with things like: long radicals, long radicals superscript, etc..

Selenium might be best for getting an accurate representation for our rendered SVG. I proposed that we places Rules in the document to be used to assist in measuring distances. Another option would be to do something similar to what KaTeX does and compare the output with LaTeX.

I'm having trouble compiling this library (proposing a solution)

I'm having trouble compiling this library.

The issue seems to be with the static_map dependency. Notably,

   Compiling font_types v0.1.0 (/Users/colbyn/Scratch/ReX/fonts/types)
   Compiling stix2 v0.1.0 (/Users/colbyn/Scratch/ReX/fonts/stix2)
error: proc-macro derive panicked
    --> fonts/stix2/src/variants.rs:1660:65
     |
1660 |   pub static HORZ_VARIANTS: static_map::Map<u32, GlyphVariants> = static_map! {
     |  _________________________________________________________________^
1661 | |     Default: GlyphVariants { constructable: None, replacements: &[] },
1662 | |     0x302 => GlyphVariants { // uni0302
1663 | |         constructable: None,
...    |
2864 | |
2865 | | };
     | |_^
     |
     = help: message: assertion failed: input.starts_with(LEADING)
     = note: this error originates in a macro (in Nightly builds, run with -Z macro-backtrace for more info)

error: proc-macro derive panicked
    --> fonts/stix2/src/variants.rs:5:65
     |
5    |   pub static VERT_VARIANTS: static_map::Map<u32, GlyphVariants> = static_map! {
     |  _________________________________________________________________^
6    | |     Default: GlyphVariants { constructable: None, replacements: &[] },
7    | |     0x2F => GlyphVariants { // slash
8    | |         constructable: None,
...    |
1656 | |
1657 | | };
     | |_^
     |
     = help: message: assertion failed: input.starts_with(LEADING)
     = note: this error originates in a macro (in Nightly builds, run with -Z macro-backtrace for more info)

error: proc-macro derive panicked
    --> fonts/stix2/src/symbols.rs:12:61
     |
12   |   pub static SYMBOLS: static_map::Map<&'static str, Symbol> = static_map! {
     |  _____________________________________________________________^
13   | |     Default: Symbol { unicode: 0x00, atom_type: AtomType::Accent },
14   | |     // unicode-math.dtx command table
15   | |     "mathexclam" => Symbol { unicode: 0x21, atom_type: AtomType::Close }, // 33
...    |
2537 | |     "|" => Symbol { unicode: 0x2016, atom_type: AtomType::Fence }, // 8214
2538 | | };
     | |_^
     |
     = help: message: assertion failed: input.starts_with(LEADING)
     = note: this error originates in a macro (in Nightly builds, run with -Z macro-backtrace for more info)

error: proc-macro derive panicked
    --> fonts/stix2/src/glyphs.rs:15:50
     |
15   |   pub static GLYPHS: static_map::Map<u32, Glyph> = static_map! {
     |  __________________________________________________^
16   | |     Default: Glyph { unicode: 0x0000_u32, bbox: BBox(fontunit!(0), fontunit!(0), fontunit!(0), fontunit!(0)), attachment: fontunit!(0), a...
17   | |
18   | |     0x0020_u32 => Glyph { unicode: 0x0020_u32, bbox: BBox(fontunit!(0), fontunit!(0), fontunit!(0), fontunit!(0)), attachment: fontunit!(...
...    |
5290 | |     0x1d455_u32 => Glyph { unicode: 0x210e_u32, bbox: BBox(fontunit!(47), fontunit!(-10), fontunit!(549), fontunit!(706)), attachment: fo...
5291 | | };
     | |_^
     |
     = help: message: assertion failed: input.starts_with(LEADING)
     = note: this error originates in a macro (in Nightly builds, run with -Z macro-backtrace for more info)

error: cannot find macro `__static_map__construct_map` in this scope
    --> fonts/stix2/src/variants.rs:1660:65
     |
1660 |   pub static HORZ_VARIANTS: static_map::Map<u32, GlyphVariants> = static_map! {
     |  _________________________________________________________________^
1661 | |     Default: GlyphVariants { constructable: None, replacements: &[] },
1662 | |     0x302 => GlyphVariants { // uni0302
1663 | |         constructable: None,
...    |
2864 | |
2865 | | };
     | |_^
     |
     = note: this error originates in a macro (in Nightly builds, run with -Z macro-backtrace for more info)

error: cannot find macro `__static_map__construct_map` in this scope
    --> fonts/stix2/src/variants.rs:5:65
     |
5    |   pub static VERT_VARIANTS: static_map::Map<u32, GlyphVariants> = static_map! {
     |  _________________________________________________________________^
6    | |     Default: GlyphVariants { constructable: None, replacements: &[] },
7    | |     0x2F => GlyphVariants { // slash
8    | |         constructable: None,
...    |
1656 | |
1657 | | };
     | |_^
     |
     = note: this error originates in a macro (in Nightly builds, run with -Z macro-backtrace for more info)

The library seems to be unmaintained.

Any interest in replacing the library with something simple?

Perhaps something akin to:

macro_rules! indexfor {
    ($keys:expr, $target:expr) => {{
        // let mut index: Option<usize> = None;
        let mut index: usize = 0;
        let mut result: Option<usize> = None;
        for k in $keys {
            if $target == *k {
                result = Some(index);
                break;
            }
            index = index + 1;
        }
        result.unwrap()
    }};
}

macro_rules! hardcoded {
    ($name:ident ($kty:ty, $vty:ty) {$($key:expr => $value:expr),* $(,)?}) => {
        #[allow(non_upper_case_globals)]
        pub static Entries: &'static [($kty, $vty)] = &[
            $(
                ($key, $value),
            )*
        ];

        macro_rules! $name {
            ($id:expr) => {{
                let index = indexfor!(&[$($key),*], $id);
                Entries[index].1
            }};
        }
    };
}

hardcoded!{
    get_symbol(&str, &str) {
        "a" => "lorem",
        "b" => "ipsum",
        "c" => "red",
    }
}



fn main() {
    let x = get_symbol!("a");
    println!("Output: {:#?}", x);
}

Instead of a Map like data structure, we just replace it with a static array, using the getter generated by hardcoded!, i.e. get_symbol!(key).

PS: Awesome library! I can see this replacing libraries like iosMath, that offer math rendering without embedding a browser.

I wish I had known about this library when I started working on Subscript for my school math notes.

Using a fixed point FontUnit dimension for internal calculations.

On IRC we discussed the possibility of using a scale of internal font units (in xits it's 1/1000 em) for internal dimension in ReX. We concluded that it would probably be best if we use a 24.8 fixedpoint representation of the internal font unit. A cursory glance of the code showed that *2, /2, and +/- is by far the most common operation. However, *0.68 is used for scripts scaling, and it is unclear what the distribution of *0.68 use is to make a fully informed decision.

Pertinent numbers from the intel optimization manual:

	Latency	Throughput
ADD	1	0.5
MUL	7,8	6,7
SHL	1	1
FADD	5	4,1
FMUL	5	1

Considering the number of *2, /2 uses, it seems unlikely that f64 will afford the same bitshift optimizations that a u32 would. FDIV is not listed, but expensive.

sebk also pointed out that it could be possible to put a scaling call in both the layout and renderer, but depending on a feature flag, one would be a no op and the other wouldn't.

Using a scaled font unit would also rid of all the Pixels -> FontUnit conversions in the Layout (there seem to be quite a few).

Support for Latin Modern Math

Hi cbreeden,

This is a continuation of the reddit thread; it'd be great to get support for OpenType working, or at least a primitive version. I tried to get Latin Modern Math working today and this is how much I've accomplished so far. It would be great to get your mentorship and progress further.

Setup

• Download the latest version of Latin Modern math.
• Clone https://github.com/ReTeX/rex-fonts/

Generating static files

• Run python3 rexify.py latinmodern-math.otf out/lmmodern in the rex-fonts directory to generate the files (e.g., glyphs.rs).

Issue. Due to changes changes in Rust 2018 and my recent PR, you need to change the glyphs.mako.rs file. I will submit a PR soon.

Integration

• Create a new crate in ReX/fonts, and copy over the files generated in the previous step. However, the styles.rs, kernings.rs and macros.rs (and lib.rs) are missing, so I just copied them over from the ReX/fonts/stix2 directory. Note that kernings.rs is supposed to be generated during the previous script, but the rexify.py reports that no kerning tables are found.

• In the root Cargo.toml file, point stix2 are the new crate (never mind the fact that it's still called "stix2").

• cargo build --release

[Bug.] In the generated styles.rs, we need to add the following lines. It's interesting that the python script doesn't generate these lines automatically, but they are somehow found in the stix2 package.

pub const NULL_DELIMITER_SPACE: FontUnit = fontunit!(100);
pub const DELIMITER_SHORT_FALL: FontUnit = fontunit!(100);
pub const DELIMITER_FACTOR: FontUnit = fontunit_raw!(0xE6); // 90.1%

• Copy over the font from the first step, and change main.rs so the generated SVG reads from the correct file.

---

Here is the SVG produced for \sum_{n\geq1} \frac{1}{n^2} = \pi^2/6, which looks like this. The problem is that operator symbols are somehow not working. I would need some assistance to figure out how to fix this -- where does the code look for these tables? (My guess is that the automatically generated symbols.rs somehow is not working.)

<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
<svg width="639.3165" height="260.053125" viewBox="0 0 6659 2708" xmlns="http://www.w3.org/2000/svg">
    <defs>
    <style type="text/css">@font-face{font-family:rex;src:url('rex-lmodern.otf');}</style>
    </defs>
    <g font-family="rex" font-size="1000">
<rect x="250" y="100" width="6159" height="1343" fill="none" stroke="blue" stroke-width="8"/>
<rect x="250" y="293" width="1444" height="2315" fill="none" stroke="red" stroke-width="8"/>
<rect x="972" y="293" width="1444" height="0" fill="none" stroke="blue" stroke-width="8"/>
<rect x="972" y="293" width="0" height="0" fill="none" stroke="blue" stroke-width="8"/>
<rect x="250" y="493" width="1444" height="950" fill="none" stroke="blue" stroke-width="8"/>
<rect x="250" y="493" width="1444" height="1400" fill="none" stroke="red" stroke-width="8"/>
<rect x="250" y="493" width="1444" height="1400" fill="none" stroke="green" stroke-width="8"/>
<text transform="translate(250, 1443)"></text>
<rect x="315" y="2060" width="1444" height="465" fill="none" stroke="blue" stroke-width="8"/>
<rect x="315" y="2060" width="1313" height="465" fill="none" stroke="blue" stroke-width="8"/>
<rect x="315" y="2216" width="419" height="316" fill="none" stroke="green" stroke-width="8"/>
<text transform="translate(315, 2525) scale(0.70)">𝑛</text>
<rect x="734" y="2084" width="543" height="524" fill="none" stroke="green" stroke-width="8"/>
<text transform="translate(734, 2525) scale(0.70)">≥</text>
<rect x="1278" y="2060" width="349" height="465" fill="none" stroke="green" stroke-width="8"/>
<text transform="translate(1278, 2525) scale(0.70)">1</text>
<rect x="1960" y="100" width="949" height="2040" fill="none" stroke="red" stroke-width="8"/>
<rect x="2185" y="100" width="949" height="666" fill="none" stroke="blue" stroke-width="8"/>
<rect x="2185" y="100" width="500" height="666" fill="none" stroke="green" stroke-width="8"/>
<text transform="translate(2185, 766)">1</text>
<rect x="1960" y ="1173" width="949" height="40" fill="#000"/>
<rect x="1960" y="1374" width="949" height="754" fill="none" stroke="blue" stroke-width="8"/>
<rect x="1960" y="1687" width="600" height="442" fill="none" stroke="blue" stroke-width="8"/>
<rect x="1960" y="1687" width="600" height="453" fill="none" stroke="green" stroke-width="8"/>
<text transform="translate(1960, 2129)">𝑛</text>
<rect x="2560" y="1374" width="349" height="754" fill="none" stroke="red" stroke-width="8"/>
<rect x="2560" y="1374" width="349" height="465" fill="none" stroke="blue" stroke-width="8"/>
<rect x="2560" y="1374" width="349" height="465" fill="none" stroke="green" stroke-width="8"/>
<text transform="translate(2560, 1840) scale(0.70)">2</text>
<rect x="3343" y="1076" width="778" height="234" fill="none" stroke="green" stroke-width="8"/>
<text transform="translate(3343, 1443)">=</text>
<rect x="4454" y="1012" width="570" height="431" fill="none" stroke="blue" stroke-width="8"/>
<rect x="4454" y="1012" width="570" height="442" fill="none" stroke="green" stroke-width="8"/>
<text transform="translate(4454, 1443)">𝜋</text>
<rect x="5024" y="614" width="384" height="828" fill="none" stroke="red" stroke-width="8"/>
<rect x="5024" y="614" width="384" height="465" fill="none" stroke="blue" stroke-width="8"/>
<rect x="5059" y="614" width="349" height="465" fill="none" stroke="green" stroke-width="8"/>
<text transform="translate(5059, 1080) scale(0.70)">2</text>
<rect x="5409" y="693" width="500" height="1000" fill="none" stroke="green" stroke-width="8"/>
<text transform="translate(5409, 1443)">/</text>
<rect x="5909" y="777" width="500" height="688" fill="none" stroke="green" stroke-width="8"/>
<text transform="translate(5909, 1443)">6</text>
</g>
</svg>

logging…

I would like to see output on the JS side looking like:

layout …  Ok
    [I] preparing XYZ
    [W] problem ..
    [I] finishing XYZ

rendering … Error
    [I] allocating resources
    [I] collecting glyphs
    drawing formula … Ok
        [I] rasterizing glyphs
        [I] painting glyphs
    [E] cannot save image

this might be created by the following calls

let layout_log = log.create_context("layout");
layout_log.info("preparing {}", z.name);
layout_log.warn("problem {:d}", …);
layout_log.info("finishing {}", z.name);

let render_log = log.create_context("rendering");
render_log.info("allocating resouces");
let drawing_log = render_log.create_context("drawing formula");
drawing_log.info("rasterizing glyphs");
drawing_log.info("painting glyphs");

render_log.error("cannot save image");

Radical symbol depth is too much

Radicals go too far down, shift them up

License files

The README.md has a license section that states the following:

See LICENSE-APACHE, and LICENSE-MIT for details.

These two files, however, are not in the repository. Could you put the respective files into the repository root such that the project is unambigously usable under the MIT and Apache 2.0 licenses? Thanks!

Implement a builder system for ParseNodes

This can be useful for

1. Making it easier to test parsing output.
2. Could provide an API for those who want to build an AST instead of parsing latex.

Accent support progress

Accent Support Progress

• If there is no accent, typeset like normal.
• Take largest accent smaller than nucleus.
• Layout nucleus with style cramped.
• Baseline of result == baseline of base.
• Determine offset of accent:
  • Accent has attachment correction:
    • If accentee has attachment correction,
      then align attachment corrections of both.
    • Otherwise, align attachment correction of
      accentee with center of nucleus, plus
      italics correction of nucleus is a symbol.
  • Accent has no attachment correction:
    • If accentee has attachment correction,
      center of accent with accent correction of base.
    • Align accent center with base center (plus)
      italics correction if it's a symbol.
• For superscripts, if accentee is simple symbol,
  scripts should not take accent into account for height. See #4.
• The width of the resulting box is the width of the base.
• BottomAccent: vertical placement is directly below nucleus,
  no correction takes place.
• WideAccent vs Accent: Don't expand non-WideAccent types by default.
• Implement FlattenedAccentBaseHeight.

fractions denominator spacing

Giving fractions tall symbols in small scales isn't working correctly.

Test \scrtipscriptstyle\frac{\int x}{\int x}.

Kerning Support for \text

This is trick. Some math fonts don't include standard kerning for the glyphs, will most likely need to resort to a different font file if desired. I noticed that KaTeX and MathJax neither support kerning for \text

Script/ScriptScriptStyle radicals have wrong height

While troubleshooting some inconsistency with radical layout, I tested:

\sqrt{\int x}\textstyle\sqrt{\int x}\scriptstyle\sqrt{\int x}\scriptscriptstyle\sqrt{\int x} 

The first two gave identical results with LuaLaTeX and Rex, but the last two are shown here:

The left is ReX and the right is LuaTeX.

Nolimits algorithm doesn't scale properly

We should be taking the italics correction from the variant operator glyph instead. We need to layout the contents of the base before taking italics corrections for base.

Update codegen for new kerning/variant types.

I changed the types for the kerning table to take a static slice over an array, since some fonts have different size requirements here.

I also changed the variant tables to use static_map!. Codegen needs to be updated accordingly.

Fail to render 'h'

A symbol for 'h' is equivalent to the planck constant 'h'. The unicode standard omits the mathematical 'h' to avoid the duplication.
ReX currently uses this non-existing glyph and fails to render 'h'.

A workaround is to check for this special case and use the alternative codepoint.

Find and replace

match codepoint {
        LOWER_A..=LOWER_Z => style_lookup(&LATIN_LOWER_LUT, codepoint, style),

with

const LOWER_H: u32 = 'h' as u32;

match codepoint {
        LOWER_H..=LOWER_H => 0x210E,  // planck constant
        LOWER_A..=LOWER_Z => style_lookup(&LATIN_LOWER_LUT, codepoint, style),

Investigate Vertical Layout Handling of Rules

It appears that I don't layout Rules the same way I layout Glyphs while in Vertical mode. This has caused me to miscalculate a few different layout algorithms that use rules (ie: fraction). Check for consistency.

Fix italics correction handling

Test with (T)(F)(P)(V)(W)(Y)

Accents don't scale correctly

Accents are too narrow in scaled variations. Accents are too high in scaled variations. Test \overbrace{1234456789}

NULL_DELIMITER_SPACE Accurate?

While testing fractions, it appears that NULL_DELIMITER_SPACE doesn't accurate represent the correct spacing. Check that the scaling is correct.

Correctly handle greek commands `\alpha`, `\beta`, etc.

Font property selection (italic, bold, monospace) should be recognized when using \alpha, \beta, etc.. But only is recognized when using directly unicode input for now. We should probably find a list of all such symbols that are commands as well.

Operator Limits Tests

This is a list of tests that I think should be added when we have a testing framework.

Operators that are symbols should have same baseline w/ or w/o scripts. Also italics correction offset.

• \displaystyle\int\limits_0^1\int
• \textstyle\int\limits_0^1\int
• \scriptstyle\int\limits_0^1\int
• \scriptscriptstyle\int\limits_0^1\int

Limits should be displayed only as appropriate (depending on \limits and display style)

• \displaystyle\int_0^1\sum_0^1
• \textstyle\int_0^1\sum_0^1
• \displaystyle\int\displaylimits_0^1
• \textstyle\int\displaylimits_0^1

Operators should be centered with respect to scripts

• \displaystyle\mathop{\mathrm{lim}}\limits_{x \to \infity}
• \displaystyle\sum_{a+b+c+d}^{a+b+c+d+e+f}
• \displaystyle\sum_k^n

Small scripts (not too hight) should have same base line

• \displaystyle\mathop{\mathrm{lim}}\limits_{x \to \infty}\mathop{\mathrm{lim}}\limits_{M \to \infty}
• \displaystyle\mathop{\mathrm{lim\,sup}}\limits_{x \to \infty}\mathop{\mathrm{lim\,sup}}\limits_{M \to \infty}

Project status?

Hey I'm interested in contributing to this project. Any words on the status of this project?

Since it seems like there is an unofficial rewrite over here that hasn't been merged back into this repo.

failed to build with rustc 1.21.0-nightly (1d2a6df38 2017-08-03)

Hello Developers,

ReX at 69593e240f10f3976ea4a86c1243678483c2de03 failed to build with
rustc 1.21.0-nightly (1d2a6df38 2017-08-03)

   Compiling rex v0.1.0 (file:///scratch/github/ReX)
error: unexpected end of macro invocation
   --> src/layout/engine.rs:578:17
    |
578 |         contents,
    |                 ^

error: Could not compile `rex`.

Thanks

Change Name to ℛℯˣ

State change in scripts

R_\mathfrak{e} should work.

The minus sign symbol is incorrect

Our minus sign symbol is much smaller

Accents should not place simple symbols in a `ParseNode::Group`

Currently, \hat{A} will place A in a ParseNode::Group, and \hat A will simply place a ParseNode::Symbol in the nucleus of the ParseNode::Accent.

As a macro argument, the opening and closing braces should be transparent to the processing of ParseNode::Accent. In particular, this will give false positives to nucleus.is_symbol when processing superscripts.

It may be worth considering changing the nucleus field of an Accent to a Vec<ParseNode> instead.

Try `dtoa` for float printing

Generating the SVG seems to be expensive, I'm curious if using a different library for displaying floats could reduce the load. See https://github.com/dtolnay/dtoa

Though, converting to a fixed-point representation (#22) is probably better.

`\color{...}` should validate color

Currently we pass a string. We also fail to expand during this phase (even though we don't support macros yet).

Prime cannot be shown

First, thank you for this fantastic work.
This is what I've been looking for for a long time.
(My dream is to have a WYSIWYM latex math input method. ReTeX can help a lot.)

I notice that a primed symbol like t' only shows t in the online demo, without the prime. Interestingly, t" is fine.

OperatorName special symbol handling

This isn't a concern yet, as we don't have an OperatorName command, since we don't support macros. But for future reference.

In OperatorName, special handling occurs for some symbols.

See http://tex.stackexchange.com/questions/84302/what-is-the-difference-of-mathop-operatorname-and-declaremathoperator for explanation of the definition. The special symbol handling is done by:

\gdef\newmcodes@{\mathcode`\'39\mathcode`\*42\mathcode`\."613A%
 \mathcode`\-45\mathcode`\/47\mathcode`\:"603A\relax}

Font integration

I am currently working on integrating font and ReX to support loading any math enabled fonts (from OTF, WOFF, WOFF2)

@cbreeden if you have time, i'd be happy to discuss it on Zulip