Contributing guidelines

NOTE: This is a temporary document we can use to write the contributing guidelines, and once it's done we can include it directly into the repository.

Code style

Automatic code formatting

Ginkgo uses ClangFormat (executable is usually named clang-format) and a custom .clang-format configuration file (mostly based on ClangFormat's Google style) to automatically format your code. Make sure you have ClangFormat set up and running properly (basically you should be able to run make format from Ginkgo's build directory) before committing anything that will end up in a pull request against ginkgo-project/ginkgo repository. In addition, you should never modify the .clang-format configuration file shipped with Ginkgo. E.g. if ClangFormat has trouble reading this file on your system, you should install a newer version of ClangFormat, and avoid commenting out parts of the configuration file at all costs.

ClangFormat is the primary tool that helps us achieve a uniform look of Ginkgo's codebase, while reducing the learning curve of potential contributors. However, ClangFormat configuration is not expressive enough to incorporate the entire coding style, so there are several additional rules that all contributed code should follow.

Note: To learn more about how ClangFormat will format your code, see existing files in Ginkgo, .clang-format configuration file shipped with Ginkgo, and ClangFormat's documentation.

Naming scheme

Filenames

Filenames use snake_case and use the following extensions:

• C++ source files: .cpp
• C++ header files: .hpp
• CUDA source files: .cu
• CUDA header files: .cuh
• CMake utility files: .cmake
• Shell scripts: .sh

Note: A C++ source/header file is considered a CUDA file if it contains CUDA code that is not guarded with #if guards that disable this code in non-CUDA compilers. I.e. if a file can be compiled by a general C++ compiler, it's not considered a CUDA file.

TODO: Finish this section.

Macros

C++ macros (both object-like and function-like macros) use CAPITAL_CASE. They have to start with GKO_ to avoid name clashes (even if they are #undef-ed in the same file!).

Variables

Variables use snake_case.

Constants

Constants use snake_case.

Functions

Functions use snake_case.

Structures and classes

Structures and classes which do not experience polymorphic behaviour (i.e. do not contain virtual methods, nor members which experience polymorphic behaviour) use snake_case.

All other structures and classes use CamelCase.

Members

All structure / class members use the same naming scheme as they would if they were not members:

• methods use the naming scheme for functions
• data members the naming scheme for variables or constants
• type members for classes / structures

Additionally, non-public data members end with an underscore (_).

Namespaces

Namespaces use snake_case.

Template parameters

Template parameters use CamelCase.

Whitespace

Spaces and tabs are handled by ClangFormat, but blank lines are only partially handled (the current configuration doesn't allow for more than 2 blank lines). Thus, contributors should be aware of the following rules for blank lines:

1. Top-level statements and statements directly within namespaces are separated with 2 blank lines. The first / last statement of a namespace is separated by two blank lines from the opening / closing brace of the namespace.

   1. exception: if the first or the last statement in the namespace is another namespace, then no blank lines are required
      example:

      namespace foo {
      
      
      struct x {
      };
      
      
      }  // namespace foo
      
      
      namespace bar {
      namespace baz {
      
      
      void f();
      
      
      }  // namespace baz
      }  // namespace bar

   2. exception: in header files whose only purpose is to declare a bunch of functions (e.g. the *_kernel.hpp files) these declarations can be separated by only 1 blank line (note: standard rules apply for all other statements that might be present in that file)

   3. exception: "related" statement can have 1 blank line between them. "Related" is not a strictly defined adjective in this sense, but is in general one of:

      1. overload of a same function,
      2. function / class template and it's specializations,
      3. macro that modifies the meaning or adds functionality to the previous / following statement.

      However, simply calling function f from function g does not imply that f and g are "related".

2. Statements within structures / classes are separated with 1 blank line. There are no blank lines betweeen the first / last statement in the structure / class.

   1. exception: there is no blank line between an access modifier (private, protected, public) and the following statement.
      example:

      class foo {
      public:
          int get_x() const noexcept { return x_; }
      
          int &get_x() noexcept { return x_; }
      
      private:
          int x_;
      };

3. Function bodies cannot have multiple consecutive blank lines, and a single blank line can only appear between two logical sections of the function.

4. Unit tests should follow the AAA pattern, and a single blank line must appear between consecutive "A" sections. No other blank lines are allowed in unit tests.

5. Enumeration definitions should have no blank lines between consecutive enumerators.

#include statement grouping

In general, all include statements should be present on the top of the file, ordered in the following groups, with two blank lines between each group:

1. Related header file (e.g. core/foo/bar.hpp included in core/foo/bar.cpp, or in the unit testcore/test/foo/bar.cpp)
2. Standard library headers (e.g. vector)
3. System third-party library headers (e.g. omp.h)
4. Local third-party library headers
5. Public Ginkgo headers
6. Private Ginkgo headers

Example: A file core/base/my_file.cpp might have an include list like this:

#include <ginkgo/core/base/my_file.hpp>


#include <algorithm>
#include <vector>
#include <tuple>


#include <omp.h>


#include "third_party/blas/cblas.hpp"
#include "third_party/lapack/lapack.hpp"


#include <ginkgo/core/base/executor.hpp>
#include <ginkgo/core/base/lin_op.hpp>
#include <ginkgo/core/base/types.hpp>


#include "core/base/my_file_kernels.hpp"

Note: ClangFormat will take care of sorting the includes alphabetically in each group.

Other Code Formatting not handled by ClangFormat

Control flow constructs

Single line statements should be avoided in all cases. Use of brackets is mandatory for all control flow constructs (e.g. if, for, while, ...).

Variable declarations

C++ supports declaring / defining multiple variables using a single type-specifier. However, this is often very confusing as references and pointers exhibit strange behavior:

template <typename T> using pointer = T *;

int *        x, y;  // x is a pointer, y is not
pointer<int> x, y;  // both x and y are pointers

For this reason, always declare each variable on a separate line, with its own type-specifier.

CMake coding style

Whitespaces

All alignment in CMake files should use four spaces.

Use of macros vs functions

Macros in CMake do not have a scope. This means that any variable set in this macro will be available to the whole project. In contrast, functions in CMake have local scope and therefore all set variables are local only. In general, wrap all piece of algorithms using temporary variables in a function and use macros to propagate variables to the whole project.

Naming style

All Ginkgo specific variables should be prefixed with a GINKGO_ and all functions by ginkgo_.

Documentation style

Documentation uses standard Doxygen.

Developer targeted notes

Make use of @internal doxygen tag. This can be used for any comment which is not intended for users, but is useful to better understand a piece of code.

Whitespaces

After named tags such as @param foo

The documentation tags which use an additional name should be followed by two spaces in order to better distinguish the text from the doxygen tag. It is also possible to use a line break instead.

Other programming comments

std::cout and other standard stream objects

These are global objects and are shared inside the same translation unit. Therefore, whenever its state or formatting is changed (e.g. using std::hex or floating point formatting) inside library code, make sure to restore the state before returning the control to the user. See this stackoverflow question for examples on how to do it correctly. This is extremely important for header files.

Project structure

Ginkgo is divided into a core module with common functionalities independent of the architecture, and several kernel modules (reference, cpu, gpu) wich contain low-level computational routines for each supported architecture.

Extended header files

Some header files from the core module have to be extended to include special functionality for specific architectures. An example of this is core/base/math.hpp, which has a GPU counterpart in gpu/base/math.hpp. For such files you should always include the version from the module you are working on, and this file will internally include its core counterpart.

Using library classes

Creating new classes, it is allowed to use existing classes (polymorphic objects) inside the kernels for the distinct backends (reference/cuda/omp...). However, it is not allowed to construct the kernels by creating new instances of a distinct (existing) class as this can result in compilation problems.

For example, when creating a new matrix class AB by combining existing classes A and B, the AB::apply() function composed of kernel invocations to A::apply() and B::apply() can only be defined in the core module, it is not possible to create instances of A and B inside the AB kernel file.