4.1 KiB
Documentation
On this page, we refer to some internal documentation rules. We use doxygen to generate our documentation and code reference. The most important conventions for documentation in CArL are collected here.
Note that some of the documentation may be incomplete or rendered incorrectly, especially if you use an old version of doxygen. Here is a list of known problems:
- Comments in code blocks (see below) may not work correctly (e.g. with doxygen 1.8.1.2). See here for a workaround. This will however look ugly for newer doxygen versions, hence we do not use it.
- Files with
static_assert
statements will be incomplete. A patch is pending and will hopefully make it into doxygen 1.8.9. - Member groups (usually used to group operators) may or may not work. There still seem to be a few cases where doxygen messes up.
- Documenting unnamed parameters is not possible. A corresponding ticket exists for several years.
Modules
In order to structure the reference, we use the concept of
Doxygen modules.
Such modules are best thought of as a hierarchical set of tags, called groups.
We define those groups in /doc/markdown/codedocs/groups.dox
.
Please make sure to put new files and classes in the appropriate groups.
Literature references
Literature references should be provided when appropriate.
We use a bibtex database located at /doc/literature.bib
with the following conventions:
- Label for one author:
LastnameYY
, for exampleDucos00
for @cite Ducos00 . - Label for multiple authors:
ABCYY
whereABC
are the first letters of the authors last names. For exampleGCL92
for @cite GCL92 . - Order the bibtex entrys by label.
These references can be used with @cite label
, for example like this:
@code
/**
- Checks whether the polynomial is unit normal
- @see @cite GCL92, page 39
- @return If polynomial is normal. */ bool isNormal() const; @endcode
Code comments
File headers
@code /**
Descriptions may be omitted when the file contains a single class, either implementation or declaration.
Namespaces
Namespaces are documented in a separate file, found at '/doc/markdown/codedocs/namespaces.dox'
Class headers
@code /**
Method headers
@code /**
- [ Usage Description ]
- @param [ Short description for first parameter ]
- @param [ Short description for second parameter ]
- @return [ Short description of return value ]
- @see
- @see */ @endcode
These method headers are written directly above the method declaration. Comments about the implementation are written above the or inside the implementation.
The see
command is used likewise as for classes.
Method groups
There are some cases when documenting each method is tedious and meaningless, for example operators. In this case, we use doxygen method groups.
For member operators (for example operator+=
), this works as follows:
@code
/// @name In-place addition operators
/// @{
/**
- Add something to this polynomial and return the changed polynomial.
- @param rhs Right hand side.
- @return Changed polynomial. */ MultivariatePolynomial& operator+=(const MultivariatePolynomial& rhs); MultivariatePolynomial& operator+=(const Term& rhs); MultivariatePolynomial& operator+=(const Monomial& rhs); MultivariatePolynomial& operator+=(Variable::Arg rhs); MultivariatePolynomial& operator+=(const Coeff& rhs); /// @}
Writing out-of-source documentation
Documentation not directly related to the source code is written in Markdown format, and is located in
/doc/markdown/
.