GCCXML
has been superseded by CastXML
. It is highly recommended to use CastXML
. GCCXML
support will be removed from Pygccxml
in version 2.0.
Pygccxml
supports C++98
, as CastXML
and GCCXML
only output declarations from the C++98
subset. Of course, newer versions of C++ can be parsed (the tests currently all pass with C++11
and C++14
), but not all new features from these language definitions can be used.
C
code support has been reported to work. As C
is similar to C++
, this makes sense. Some discrepancies may be present.
Still, parsing C
code is not officially supported by pygccxml
, as it falls out of scope of this project. Of course, if some volunteer wants to work on this, submissions would be accepted.
Pygccxml
does not allow to fetch declarations defined in function or method bodies. For example the following a
variable will not appear in the declarations tree:
int f() {
int a = 3;
return a;
}
Neither GCCXML
or CastXML
currently support this feature. CastXML
could probably be extended for this later, as pygccxml
.
pygccxml is being regularly optimised for performance, but it still may be slow in certain cases.
Before all, it is highly recommended to benchmark your application if performance is important to you. There are multiple tools out there for benchmarking python applications. We currently are using the following two command lines / tools:
python -m cProfile -o profile_data.pyprof script_to_profile.py
pyprof2calltree -i profile_data.pyprof -k
Of course optimising pygccxml alone will not help in all cases. The bottlenecks can also be in the code calling pygccxml, to make sure to benchmark the whole process. Any help on the performance side is also welcome.
Some things you may try (in order of priority):
You might want to consider making the declaration tree as small as possible and only store those declarations that somehow have an influence on the bindings. Ideally, this is done as early as possible and luckily castxml and gccxml provide an option that allows you to reduce the number of declarations that need to be parsed.
You can specify one or more declarations using the
-fxml-start
(gccxml) or-castxml-start
(castxml) options when running the xml generator. For example, if you specify the name of a particular class, only this class and all its members will get written. Ideally, your project should already use a dedicated namespace, which you can then use as a starting point. All declarations stemming from system headers will be ignored (except for those declarations that are actually used within your library).In the pygccxml package you can set the value for these flags by using the
start_with_declarations
attribute of thepygccxml.parser.config_t
object that you are passing to the parser.You can pass the following flag to the read_files method:
compilation_mode=pygccxml.parser.COMPILATION_MODE.ALL_AT_ONCE
- If you want to cache the declarations tree, there is a caching mechanism provided by pygccxml. You will find an example of this mechanism in the examples section.
The `castxml_epic_version
` can be set to 1 to benefit from new castxml and pygccxml features. To be able to use this, you will need the latest castxml version.
Currently this adds the support for elaborated type specifiers.
__va_list_tag and other hidden declarations (f1)
When parsing with CastXML, the XML tree can contain declarations named __va_list_tag
. If the compiler is llvm 3.9, __NSConstantString_tag
and __NSConstantString
declarations may also be present.
These declarations are internal declarations, coming from the std c++ library headers you include, and are often not needed. They are for example polluting the declarations tree when running pyplusplus.
By default, pygccxml will ignore these declarations. To still read these declarations from the xml file, a config flag can be set (config.flags = ["f1"]
), or a flag can be passed as argument the config setup (flags=["f1"]
).
Attributes defined as `__thiscall__
are now ignored (tested with VS 2013). The __thiscall__
in some attributes will be removed too. If you still want to have access to these attributes, you can use the config.flags = ["f2"]` option.