1 Allowed language features
2 =========================
4 Most of these are not strict rules, but you should have a very good
5 reason for deviating from them.
7 Portability considerations
8 ^^^^^^^^^^^^^^^^^^^^^^^^^^
10 Most |Gromacs| files compile as C++14, but some files remain that compile as C99.
11 C++ has a lot of features, but to keep the source code maintainable and easy to read,
12 we will avoid using some of them in |Gromacs| code. The basic principle is to keep things
13 as simple as possible.
14 For compatiblity, certain work-arounds are required because not all compilers support
15 these standards fully.
17 * MSVC supports only a subset of C99 and work-arounds are required in those cases.
18 * We should be able to use virtually all C++11 features outside of OpenCL kernels
19 (which compile as C), and for consistency also in CUDA kernels.
24 |Gromacs| code must support the lowest common denominator of C++14 standard library
25 features available on supported platforms.
26 Some modern features are useful enough to warrant back-porting.
27 Consistent and forward-compatible headers are provided in ``src/gromacs/compat/``
28 as described in the `Library documentation <../doxygen/html-lib/group__group__compatibility.xhtml>`_
30 General considerations
31 ^^^^^^^^^^^^^^^^^^^^^^
32 As a baseline, |Gromacs| follows the C++ Core Guidelines |linkref1|, unless
33 our own more specific guidelines below say otherwise. We tend to be more restrictive
34 in some areas, both because we depend on the code compiling with a lot of different
35 C++ compilers, and because we want to increase readability. However, |Gromacs| is an
36 advanced projects in constant development, and as our needs evolve we will both
37 relax and tighten many of these points. Some of these changes happen naturally as
38 part of agreements in code review, while major parts where we don't agree should be
39 pushed to a redmine thread. Large changes should be suggested early in the development
40 cycle for each release so we avoid being hit by last-minute compiler bugs just before
43 * Use namespaces. Everything in ``libgromacs`` should be in a ``gmx``
44 namespace. Don't use using in headers except possibly for aliasing
45 some commonly-used names, and avoid file-level blanket ``using
46 namespace gmx`` and similar. If only a small number of ``gmx``
47 namespace symbols needed in a not-yet-updated file, consider
48 importing just those symbols. See also |linkref2|.
49 * Use STL, but do not use iostreams outside of the unit tests. iostreams can have
50 a negative impact on performance compared to other forms
51 of string streams, depending on the use case. Also, they don't always
52 play well with using C ``stdio`` routines at the same time, which
53 are used extensively in the current code. However, since Google tests
54 rely on iostreams, you should use it in the unit test code.
55 * Don't use non-const references as function parameters. They make it
56 impossible to tell whether a variable passed as a parameter may
57 change as a result of a function call without looking up the
59 * Use ``not_null<T>`` pointers wherever possible to convey the
60 semantics that a pointer to a valid is required, and a reference
61 is inappropriate. See also |linkrefnotnull|.
62 * Don't use C-style casts; use ``const_cast``, ``static_cast`` or
63 ``reinterpret_cast as appropriate``. See the point on RTTI for
64 ``dynamic_cast``. For emphasizing type (e.g. intentional integer division)
65 use constructor syntax. For creating real constants use the user-defined literal
66 _real (e.g. 2.5_real instead of static_cast<real>(2.5)).
67 * Avoid overloading functions unless all variants really do the same
68 thing, just with different types. Instead, consider making the
69 function names more descriptive.
70 * Avoid using default function arguments. They can lead to the code
71 being less readable than without (see |linkref3|). If you think that your specific
72 case improves readability (see |linkref4|), you can justify their use.
73 * Don't overload operators before thorough consideration whether it
74 really is the best thing to do. Never overload ``&&``, ``||``, or
75 the comma operator, because it's impossible to keep their original
76 behavior with respect to evaluation order.
77 * Try to avoid complex templates, complex template specialization or
78 techniques like SFINAE as much as possible. If nothing else, they
79 can make the code more difficult to understand.
80 * Don't use multiple inheritance. Inheriting from multiple pure
81 interfaces is OK, as long as at most one base class (which should be
82 the first base class) has any code. Please also refer to the
83 explanation |linkref5| and |linkref6|.
84 * Don't write excessively deep inheritance graphs. Try to not inherit
85 implementation just to save a bit of coding; follow the principle
86 "inherit to be reused, not to reuse." Also, you should not
87 mix implementation and interface inheritance. For explanation please
89 * Don't include unnecessary headers.
90 * Make liberal use of assertions to help document your intentions (but
91 prefer to write the code such that no assertion is necessary).
92 * Prefer ``GMX_ASSERT()`` and ``GMX_RELEASE_ASSERT()`` to naked
93 ``assert()`` because the former permit you to add descriptive text.
94 * Use gmx::Mutex rather than pthreads, std or raw thread-MPI mutexes.
95 * Use proper enums for variable whose type can only contain one of a
96 limited set of values. C++ is much better than C in catching errors
97 in such code. Ideally, all enums should be typed enums, please
99 * When writing a new class, think whether it will be necessary to make
100 copies of that class. If not, declare the copy constructor and the
101 assignment operator as private and don't define them, making any
102 attempt to copy objects of that class fail. If you allow copies,
103 either provide the copy constructor and the assignment operator, or
104 write a clear comment that the compiler-generated ones will do (and
105 make sure that they do what you
106 want). ``src/gromacs/utility/classhelpers.h`` has some convenience
107 macros for doing this well.
108 You can also use deleted functions in this case.
109 * Declare all constructors with one parameter as explicit unless you
110 really know what you are doing. Otherwise, they can be used for
111 implicit type conversions, which can make the code difficult to
112 understand, or even hide bugs that would be otherwise reported by
113 the compiler. For the same reason, don't declare operators for
114 converting your classes to other types without thorough
115 consideration. For an explanation, please see |linkref9|.
116 * Write const-correct code (no ``const_cast`` unless absolutely
118 * Avoid using RTTI (run-time type information, in practice
119 ``dynamic_cast`` and ``typeid``) unless you really need it. The cost
120 of RTTI is very high, both in binary size (which you always
121 pay if you compile with it) and in execution time (which you pay
122 only if you use it). If your problem seems to require RTTI, think
123 about whether there would be an alternative design that
124 wouldn't. Such alternative designs are often better.
125 * Don't depend on compiler metadata propagation. struct elements
126 and captured lambda parameters tend to have ``restrict`` and
127 alignment qualifiers discarded by compilers, so when you later
128 define an instance of that structure or allocate memory to
129 hold it, the data member might not be aligned at all.
130 * Plan for code that runs in compute-sensitive kernels to have useful
131 data layout for re-use, alignment for SIMD memory operations
132 * Recognize that some parts of the code have different requirements -
133 compute kernels, mdrun setup code, high-level MD-loop code,
134 simulation setup tools, and analysis tools have different needs, and
135 the trade-off point between correctness vs reviewer time vs
136 developer time vs compile time vs run time will differ.
139 .. |linkref1| replace:: `c++ guidelines <http://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines>`__
140 .. |linkref2| replace:: `here <http://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines#sf7-dont-write-using-namespace-in-a-header-file>`__
141 .. |linkref3| replace:: `here <http://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines#i23-keep-the-number-of-function-arguments-low>`__
142 .. |linkref4| replace:: `here <https://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines#f51-where-there-is-a-choice-prefer-default-arguments-over-overloading>`__
143 .. |linkref5| replace:: `here <http://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines#c135-use-multiple-inheritance-to-represent-multiple-distinct-interfaces>`__
144 .. |linkref6| replace:: `here <http://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines#c136-use-multiple-inheritance-to-represent-the-union-of-implementation-attributes>`__
145 .. |linkref7| replace:: `here <http://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines#c129-when-designing-a-class-hierarchy-distinguish-between-implementation-inheritance-and-interface-inheritance>`__
146 .. |linkref8| replace:: `here <http://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines#Renum-class>`__
147 .. |linkref9| replace:: `here <http://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines#Rc-explicit>`__
148 .. |linkrefnotnull| replace:: `here <http://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines#Ri-nullptr> and here <http://isocpp.github.io/CppCoreGuidelines/CppCoreGuidelines#Rf-nullptr>`__
150 .. _implementing exceptions:
152 Implementing exceptions for error handling
153 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
154 See :ref:`error handling` for the approach to handling run-time
155 errors, ie. use exceptions.
157 * Write exception-safe code. All new code has to offer at least the
158 basic or nothrow guarantee to make this feasible.
159 * Use std (or custom) containers wherever possible.
160 * Use smart pointers for memory management. By default, use
161 ``std::unique_ptr`` and ``gmx::unique_cptr`` in assocation with any
162 necessary raw ``new`` or ``snew`` calls. ``std::shared_ptr`` can be
163 used wherever responsibility for lifetime must be shared.
164 Never use ``malloc``.
165 * Use RAII for managing resources (memory, mutexes, file handles, ...).
166 * It is preferable to avoid calling a function which might throw an
167 exception from a legacy function which is not exception safe. However,
168 we make the practical exception to permit the use of features such
169 as ``std::vector`` and ``std::string`` that could throw
170 ``std::bad_alloc`` when out of memory. In particular, |Gromacs| has
171 a lot of old C-style memory handling that checking tools continue
172 to issue valid warnings about as the tools acquire more
173 functionality, and fixing these with old constructs is an
174 inefficient use of developer time.
175 * Functions / methods should be commented whether they are exception
176 safe, whether they might throw an exception (even indirectly), and
177 if so, which exception(s) they might throw.
179 Preprocessor considerations
180 ^^^^^^^^^^^^^^^^^^^^^^^^^^^
181 * Don't use preprocessor defines for things other than directly
182 related to configuring the build. Use templates or inline functions
183 to generate code, and enums or const variables for constants.
184 * Preprocessing variables used for configuring the build should be
185 organized so that a valid value is always defined, i.e. we never
186 test whether one of our preprocessor variables is defined, rather we
187 test what value it has. This is much more robust under maintance,
188 because a compiler can tell you that the variable is undefined.
189 * Avoid code with lengthy segments whose compilation depends on #if
191 * Prefer to organize the definition of a const variable at the top of
192 the source code file, and use that in the code. This helps keep all
193 compilation paths built in all configurations, which reduces the
194 incidence of silent bugs.
195 * Indent nested preprocessor conditions if nesting is necessary and
196 the result looks clearer than without indenting.
197 * Please strongly consider a comment repeating the preprocessor condition at the end
198 of the region, if a lengthy region is neccessary and benefits from
199 that. For long regions this greatly helps in understanding
200 and debugging the code.