Framework for Writing C++ Committee Papers

Introduction

This is a paper-writing framework designed to ease the pain of authoring papers for WG21, built on top of Pandoc.

In short, you write your papers in Markdown and the framework produces the paper either in HTML or PDF.

Requirements

• python3
• xelatex (Only for PDF papers)

OS X

brew install mactex # Only for PDF papers

Ubuntu

sudo apt-get install texlive-latex-base # Only for PDF papers

Debian

Debian installation may require these additional packages:

• texlive-fonts-recommended
• texlive-latex-recommended
• texlive-latex-extra

Integration

git submodule add https://github.com/mpark/wg21.git

echo "include wg21/Makefile" > Makefile

make <paper>.pdf  # `<paper>.md` -> `generated/<paper>.pdf`
make <paper>.html # `<paper>.md` -> `generated/<paper>.html`

See mpark/wg21-papers for an example use of this project.

Generation

make <paper>.pdf  # `<paper>.md` -> `generated/<paper>.pdf`
make <paper>.html # `<paper>.md` -> `generated/<paper>.html`

Formatting

This framework provides support for various common elements for C++ papers.

• Title
• Table of Contents
• Markdown
• Embedded Markdown within Code
• Comparison Tables
• Proposed Wording
  • Paragraph Numbers
  • Code Changes
  • Wording Changes
  • Grammar Changes
• Stable Names
• Citations
• References
  • Automatic References
  • Manual References
• Unicode Considerations

Title

The title is specified in a YAML metadata block.

date: today will generate today's date in YYYY-MM-DD (ISO 8601) format.

YAML lists can be used to specify multiple audiences and authors:

---
title: Integration of chrono with text formatting
document: P1361R0
date: 2018-10-16
audience:
  - Library Evolution Working Group
  - Library Working Group
author:
  - name: Victor Zverovich
    email: <[email protected]>
  - name: Daniela Engert
    email: <[email protected]>
  - name: Howard E. Hinnant
    email: <[email protected]>
---

Table of Contents

---

---

The default toc-depth is 3, but it can be specified to go deeper:

Markdown

Refer to the full Pandoc Markdown spec for useful extensions!

Embedded Markdown within Code

Within default, cpp, and diff code elements, any text surrounded by the @ symbol is formatted as Markdown! This is useful for conventions such as see below, unspecified, and exposition-only variable names.

This also works for inline code, e.g.,

Recall the `static_cast`{.cpp} syntax: `static_cast < @_type-id_@ > ( @_expression_@ )`{.cpp}.

If you need to nest embedded Markdown, surround the outer context with @@. This comes up sometimes if you want to produce inline diffs within a code block, and some of the inner code need to be marked up.

```
template <@[`invocable`]{.rm}[`class`]{.add}@ F@[`, class`]{.add}@>
struct @_as-receiver_@ {
@[`private:`]{.rm}@
  @[`using invocable_type = std::remove_cvref_t<F>;`]{.rm}@
  @[`invocable_type`]{.rm}[`F`]{.add}@ f_;
@[`public:`]{.rm}@
  @@[`explicit @_as-receiver_@(invocable_type&& f)`]{.rm}@@
  @@[`@_as-receiver_@(@_as-receiver_@&& other) = default;`]{.rm}@@
  void set_value() @[`noexcept(is_nothrow_invocable_v<F&>)`]{.add}@ {
    invoke(f_);
  }
  @[`[[noreturn]]`]{.add}@ void set_error(std::exception_ptr) @[`noexcept`]{.add}@ {
    terminate();
  }
  void set_done() noexcept {}
};
```

Comparison Tables

Comparison Tables are fenced Div blocks that open with ::: cmptable and close with :::. Fenced code blocks are the only elements that actually get added to Comparison Tables, except that the last header (if any) before a fenced code block is attached to the cell above.

::: cmptable

### Before
```cpp
switch (x) {
  case 0: std::cout << "got zero"; break;
  case 1: std::cout << "got one"; break;
  default: std::cout << "don't care";
}
```

### After
```cpp
inspect (x) {
  0: std::cout << "got zero";
  1: std::cout << "got one";
  _: std::cout << "don't care";
}
```

:::

Each fenced code block is pushed onto the current row, and horizontal rules (---) are used to move to the next row.

::: cmptable

### Before
```cpp
switch (x) {
  case 0: std::cout << "got zero"; break;
  case 1: std::cout << "got one"; break;
  default: std::cout << "don't care";
}
```

### After
```cpp
inspect (x) {
  0: std::cout << "got zero";
  1: std::cout << "got one";
  _: std::cout << "don't care";
}
```

---

```cpp
if (s == "foo") {
  std::cout << "got foo";
} else if (s == "bar") {
  std::cout << "got bar";
} else {
  std::cout << "don't care";
}
```

```cpp
inspect (s) {
  "foo": std::cout << "got foo";
  "bar": std::cout << "got bar";
  _: std::cout << "don't care";
}
```

:::

The last block quote > caption (if any) is used as the caption.

::: cmptable

> Put your caption here

### Before
```cpp
switch (x) {
  case 0: std::cout << "got zero"; break;
  case 1: std::cout << "got one"; break;
  default: std::cout << "don't care";
}
```

### After
```cpp
inspect (x) {
  0: std::cout << "got zero";
  1: std::cout << "got one";
  _: std::cout << "don't care";
}
```

:::

Proposed Wording

Paragraph Numbers

Paragraph numbers are bracketed Span elements that look like: [2]{.pnum} and [2.1]{.pnum}.

[2]{.pnum} An expression is _potentially evaluated_ unless it is an unevaluated
operand (7.2) or a subexpression thereof. The set of _potential results_ of
an expression `e` is defined as follows:

- [2.1]{.pnum} If `e` is an _id-expression_ (7.5.4), the set contains only `e`.
- [2.2]{.pnum} If `e` is a subscripting operation (7.6.1.1) with an array operand,
the set contains the potential results of that operand.

Code Changes

Wording Changes

Large changes are fenced Div blocks with ::: add for additions, ::: rm for removals.

Small, inline changes are bracketed Span elements that looks like [new text]{.add} or [old text]{.rm}.

Grammar Changes

Use line blocks (|) in order to preserve the leading spaces.

> | _selection-statement:_
> |     `if constexpr`_~opt~_ `(` _init-statement~opt~_ _condition_ `)` _statement_
> |     `if constexpr`_~opt~_ `(` _init-statement~opt~_ _condition_ `)` _statement_ `else` _statement_
> |     `switch (` _init-statement~opt~_ _condition_ `)` _statement_
> |     [`inspect` `constexpr`~_opt_~ `(` _init-statement~opt~_ _condition_ `)` `{`
>            _inspect-case-seq_
>        `}`]{.add}
>
> ::: add
> | _inspect-case-seq:_
> |     _inspect-case_
> |     _inspect-case-seq_ _inspect-case_
>
> | _inspect-case:_
> |     _attribute-specifier-seq~opt~_ _inspect-pattern_ _inspect-guard~opt~_ `:` _statement_
>
> | _inspect-pattern:_
> |     _wildcard-pattern_
> |     _identifier-pattern_
> |     _constant-pattern_
> |     _structured-binding-pattern_
> |     _alternative-pattern_
> |     _binding-pattern_
> |     _extractor-pattern_
>
> | _inspect-guard:_
> |     `if (` _expression_ `)`
> :::

Stable Names

Stable names are bracketed Span elements that look like: [stable.name]{.sref}.

In [expr.sizeof]{.sref}/5:

The identifier in a `sizeof...` expression shall name a [parameter]{.rm} pack.
The `sizeof...` operator yields the number of [arguments provided for]{.rm}
[elements in]{.add} the [parameter]{.rm} pack [identifier]{.rm} ([temp.variadic]{.sref}).

You may can also add a class - or .unnumbered to omit the section number.

For example, [expr.sizeof]{- .sref} or [expr.sizeof]{.unnumbered .sref}

Run make update to update the local cache of annex-f.

Citations

In-text citations look like this: [@id]

References

Automatic References

The bibliography is automatically generated from https://wg21.link/index.yaml for citations of the following types.

Type	Id
Paper	Nxxxx / PxxxxRn
Issue	CWGxxxx / EWGxxxx / LWGxxxx / LEWGxxxx / FSxxxx
Editorial	EDITxxx
Standing Document	SDx

The [@N3546] example from Citations generates:

Run make update to update the local cache of index.yaml.

Manual References

Manual references are specified in a YAML metadata block similar to Title, typically at the bottom of the document.

The `id` field is for in-text citations (e.g., [@PAT]),
and `citation-label` is the label for the reference.

Typically `id` and `citation-label` are kept the same.

---
references:
  - id: PAT
    citation-label: Patterns
    title: "Pattern Matching in C++"
    author:
      - family: Park
        given: Michael
    URL: https://github.com/mpark/patterns
---

Unicode Considerations

If you build for LaTeX output and you have Unicode characters in any of your paper's source code, you may have problems. First, the default PDF engine simply does not support Unicode characters. You can add --pdf-engine=xelatex to the call to pandoc in the Makefile to use xelatex instead. That gives you access to some font selections for different parts of your paper (see the Fonts section of the Pandoc manual). The option that controls your source code fonts is monofont. You can add a line with your monofont choice to your YAML metadata block. Here, it's "DejaVu Sans Mono" which provides glyphs for a large amount of the Unicode characters:

---
title: Integration of chrono with text formatting
document: P1361R0
date: 2018-10-16
audience:
  - Library Evolution Working Group
  - Library Working Group
author:
  - name: Victor Zverovich
    email: <[email protected]>
  - name: Daniela Engert
    email: <[email protected]>
  - name: Howard E. Hinnant
    email: <[email protected]>
monofont: "DejaVu Sans Mono"
---

If you want the list of available fonts on your system, most supported systems will produce a list via the command-line tool fc-list.

Other Papers

• P1361: Integration of chrono with text formatting
• P1390: Suggested Reflection TS NB Resolutions

License

Distributed under the Boost Software License, Version 1.0.

Resources

• Blog Post: How I format my C++ papers
• Lightning Talk @ C++Now 2019: WG21 Paper in Markdown