The low-level API for setting static member variables stopped working
when the fastunpack option was turned on by default. The PyMethodDef
setup requires METH_O, not METH_VARARGS with fastunpack.
* REnums2018:
Fix R return by C++11 const ref enum classes
Remove unused code in r.cxx
extra doc on anonymous enums
ENH:
FIX: references to enums now functioning
DOC: Extended documentation on enumeration support in R
FIX: Corrected path to output from R tests
Reformat comments in R module
ENH: Run test for enum_thorough in R
Code style changes post review
ENH: R Module: Enumerations with values set by calls to C code, allowing arbitarily complex value expressions.
Setting enum values with calls to the C code.
* std_unordered_map_args_fix:
Include all template parameters for std_unordered_multimap and std_unordered_map
Include all template parameters for std_unordered_map macro
These implementations are not optimized, i.e. are done in a naive way in
C#, rather than using C++ functions more efficiently, but are better
than nothing.
Create new Lib/csharp/std_set.i based on the existing std_map.i and run
li_std_set unit test for C# as well.
Notice that the set operations defined by the base ISet<> interface are
not implemented yet.
There is some consolidation work to be done. The
core of the change is getRClassName2, which will
probably eventually replace getRClassName.
getRClassName seems to be in a funny state, with
a middle argument that is commented out and never
used.
My next step is to verify whether the new version
can replace it.
This test exercises many complex cases for enumerations in R.
References to enums currently fail and will be addressed in
the next patch.
Anonymous enums are not supported.
There are some strange corner cases that could be seen as
doing strange things in R. enums are allowed to have the
same underlying integer representation on the C side. This
means that the R structure ends up with two names for the
same integer, which in turn means that the reverse lookup
isn't unique. I don't think this matters. I'll add some
of these cases to the documentation.
arbitarily complex value expressions.
R-side initialization of the enumeration structure employs
delayedAssign, which allows the initialization code to be
sourced before loading the shared library, which is the
usual behaviour for code in an R package.
nested_in_template_wrap.cxx(247): warning C4244: 'initializing': conversion from 'double' to 'int', possible loss of data
python_pybuffer_wrap.cxx(2788): warning C4267: 'return': conversion from 'size_t' to 'int', possible loss of data
Modules\python.cxx(2227) : warning C4800: 'int' : forcing value to bool 'true' or 'false' (performance warning)
* ZackerySpytz-director-classes-final-methods:
Warning tweaks for destructors that are final in director classes
Documentation for directors and virtual final methods
Fixes for final destructors in director classes
Warning fix for final destructor in directors
Remove a useless warning filter
Fix the handling of director classes with final methods
The std::vector wrappers have been changed to work by default for elements that are
not default insertable, i.e. have no default constructor. This has been achieved by
not wrapping:
vector(size_type n);
Previously the above had to be ignored via %ignore.
If the above constructor is still required it can be added back in again via %extend:
%extend std::vector {
vector(size_type count) { return new std::vector< T >(count); }
}
Alternatively, the following wrapped constructor could be used as it provides near-enough
equivalent functionality:
vector(jint count, const value_type& value);
The equivalent change to std::list has also been made (std::list
wrappers were not in the previous release [3.0.12] though).
Fixes wrapping overloaded functions/constructors where a vararg
function is declared after a non-vararg function.
This is a long standing bug in the Python layer exposed since fastunpack
was turned on by default.
