cpp11 now requires R >=4.0.0, in line with the tidyverse version policy (#411).
Because cpp11 now requires R >=4.0.0, a number of previously optional tools are now always available, allowing us to remove some dead code. In particular:
R_UnwindProtect() is always available, so the defines HAS_UNWIND_PROTECT
and CPP11_UNWIND are no longer useful.
ALTREP is always available, so the file cpp11/altrep.hpp and the define
HAS_ALTREP are no longer useful.
We would like to remove the dead code regarding these tools in the future, so we ask that you please remove usage of them from your own packages (#411).
R_NO_REMAP and STRICT_R_HEADERS are now conditionally defined only if they
have not already been defined elsewhere. This is motivated by the fact that
R_NO_REMAP is becoming the default for C++ code in R 4.5.0 (#410).
Fixed a small protection issue flagged by rchk (#408).
Removed usage of the following R non-API functions:
SETLENGTH()
SET_TRUELENGTH()
SET_GROWABLE_BIT()
These functions were used as part of the efficient growable vectors that
cpp11 offered, i.e. what happens under the hood when you use push_back().
The removal of these non-API functions means that cpp11 writable vectors that
have been pushed to with push_back() will likely force 1 extra allocation
when the conversion from cpp11::writable::r_vector<T> to SEXP occurs
(typically when you return a result back to R). This does not affect the
performance of push_back() itself, and in general these growable vectors
are still quite efficient (#362).
The environment class no longer uses the non-API function
Rf_findVarInFrame3() (#367).
The exists() method now uses the new R_existsVarInFrame() function.
The SEXP conversion operator now uses the new R_getVar() function. Note
that this is stricter than Rf_findVarInFrame3() in 3 ways. The object
must exist in the environment (i.e. R_UnboundValue is no longer returned),
the object cannot be R_MissingArg, and if the object was a promise, that
promise is now evaluated. We have backported this new strictness to older
versions of R as well.
cpp11::writable::r_vector<T>::proxy now implements copy assignment.
Practically this means that x[i] = y[i] now works when both x and y
are writable vectors (#300, #339).
New writable::data_frame constructor that also takes the number of rows as
input. This accounts for the edge case where the input list has 0 columns but
you'd still like to specify a known number of rows (#272).
std::max_element() can now be used with writable vectors (#334).
Read only r_vectors now have a move constructor and move assignment
operator (#365).
Repeated assignment to a cpp11::writable::strings vector through either
x[i] = elt or x.push_back(elt) is now more performant, at the tradeoff
of slightly less safety (as long as elt is actually a CHARSXP and i is
within bounds, there is no chance of failure, which are the same kind of
invariants placed on the other vector types) (#378).
Constructors for writable vectors from initializer_list<named_arg> now
check that named_arg contains a length 1 object of the correct type, and
throws either a cpp11::type_error or std::length_error if that is not the
case (#382).
cpp11::package now errors if given a package name that hasn't been loaded
yet. Previously it would cause R to hang indefinitely (#317).
cpp11::function now protects its underlying function, for maximum safety
(#294).
cpp11::writable::r_vector<T>::iterator no longer implicitly deletes its
copy assignment operator (#360).
Added the missing implementation for x.at("name") for read only vectors
(#370).
Fixed an issue with the writable::matrix copy constructor where the
underlying SEXP should have been copied but was not. It is now consistent with
the behavior of the equivalent writable::r_vector copy constructor.
Fixed a memory leak with the cpp11::writable::r_vector move assignment
operator (#338).
Fixed an issue where writable vectors were being protected twice (#365).
The approach for the protection list managed by cpp11 has been tweaked slightly. In 0.4.6, we changed to an approach that creates one protection list per compilation unit, but we now believe we've found an approach that is guaranteed by the C++ standard to create one protection list per package, which makes slightly more sense and still has all the benefits of the reduced maintanence burden mentioned in the 0.4.6 news bullet (#364).
A side effect of this new approach is that the preserved object exposed
through protect.hpp no longer exists. We don't believe that anyone was using
this. This also means you should no longer see "unused variable" warnings
about preserved (#249).
R >=3.6.0 is now required. This is in line with (and even goes beyond) the tidyverse standard of supporting the previous 5 minor releases of R.
Implicit conversion from sexp to bool, size_t, and double has been
marked as deprecated and will be removed in the next version of cpp11. The 3
packages that were using this have been notified and sent PRs. The recommended
approach is to instead use cpp11::as_cpp<T>, which performs type and length
checking, making it much safer to use.
Dropped support for gcc 4.8, mainly an issue for extremely old CentOS 7 systems which used that as their default compiler. As of June 2024, CentOS 7 is past its vendor end of support date and therefore also out of scope for Posit at this time (#359).
R >=3.5.0 is now required to use cpp11. This is in line with (and even goes
beyond) the tidyverse standard of supporting the previous 5 minor releases of
R. It also ensures that R_UnwindProtect() is available to avoid C++ memory
leaks (#332).
cpp11::preserved.release_all() has been removed. This was intended to
support expert developers on R <3.5.0 when cpp11 used a global protection
list. Since cpp11 no longer uses a global protection list and requires R
=3.5.0, it is no longer needed. As far as we can tell, no package was actively using this (#332).
cpp11 now creates one protection list per compilation unit, rather than one global protection list shared across compilation units and across packages. This greatly reduces the complexity of managing the protection list state and should make it easier to make changes to the protection list structure in the future without breaking packages compiled with older versions of cpp11 (#330).
Nested calls to cpp11::unwind_protect() are no longer supported or
encouraged. Previously, this was something that could be done for performance
improvements, but ultimately this feature has proven to cause more problems
than it is worth and is very hard to use safely. For more information, see the
new vignette("FAQ") section titled "Should I call cpp11::unwind_protect()
manually?" (#327).
The features and bug fixes from cpp11 0.4.4 have been added back in.
Davis Vaughan is now the maintainer.
as_doubles() and as_integers() now propagate missing values correctly
(#265, #319).
Fixed a performance issue related to nested unwind_protect() calls (#298).
Minor performance improvements to the cpp11 protect code. (@kevinushey)
cpp_register() gains an argument extension= governing the file extension of
the src/cpp11 file. By default it's .cpp, but .cc is now supported
as well (#292, @MichaelChirico)
Modernized the GitHub Actions workflows and updated some internal tests to better align with changes in those workflows and the latest version of R (#279).
cpp_source() errors on non-existent file (#261).
cpp_register() is quiet by default when R is non interactive (#289).
updated test to adapt to changes in R 4.2.1 (#290).
cpp11::messages() cpp11::stop() and cpp11::warning().
Set the CPP11_USE_FMT macro to use this feature in your package. (@sbearrows, #169, #208)as_double() and as_integer() methods to coerce integers to doubles and doubles to integers to doubles (@sbearrows, #46)cpp11::matrix iterators can now be used either row-wise or column-wise (the default) depending on the user's choice (@alyst, #229)writable::r_vector default constructors now return a 0 length vector when converted to SEXP (#166)r_vector constructors now disallow implicit construction with named arguments (#237)r_vector.attr() methods now return const objects, so it is a compile time error to try to assign to them (#237)+ and += operators of r_vector::[const_]iterator to conform the iterators concept:
+= updates the iterator, and + returns the updated copy, while keeping the original unchanged (@alyst, #231)cpp11::sexps (#224).Call calls in cpp11.cpp file (@sbearrows, #170)cpp_register() now includes attribute_visible in the init function, so packages compiled with C_VISIBILITY will find the init function.cpp_source() on the same file more than once (@sbearrows, #202)cpp11::linking_to (@sbearrows, #193)cpp11::stop() and replaced with C++ exceptions (@sbearrows, #203)x.empty() method to check if a vector is empty (@sbearrows, #182)x.named() method to check if a vector is named (@sbearrows, #186)na() free function to return the NA sentinels for R objects (@sbearrows, #179)tools::package_native_routine_registration_skeleton() (@sbearrows, #171)cpp11::unwind_protect() and cpp11::safe (#154)cpp_source() now gets an argument dir to allow customized temporary directory to store generated source files.
It makes it easier to debug C++ source files in non-package project via source mapping. (@renkun-ken, #156)cpp_register() now uses symbols exclusively in the .Call() interface. This allows it to be more robust in interactive use with the pkgload package.cpp_source() gains a cxx_std argument to control which C++ standard is used.
This allows you to use code from C++14 and later standards with cpp_source(). (#100)cxx_std chunk option to control the C++ standard used.cpp_source() now has much more informative error messages when compilation fails (#125, #139)cpp_source() now uses a unique name for the DLL, so works when run multiple times on the same source file on Windows (#143)writable::list_of<T> now supports modification of vectors as intended (#131).tools::package_native_routine_registration_skeleton() are no longer swallowed (#134)cpp_source() can now accept a source file called cpp11.cpp (#133)named_arg now explicitly protect their values, avoiding protection issues when using large inputs. tidyverse/readr#1145r_string(std::string) now uses Rf_mkCharLenCE() instead of Rf_mkChar(), which avoids the performance cost of checking the string length.r_vector::const_iterator::operator* is now a const method (#113, @bkietz, @xhochy)r_bool added as an adapter between bool and Rboolean values (#57, @bkietz)
data_frame() objects now have the number of rows correctly set as real length, not the reserved length (#91)
Fixed potential memory leak in cpp11::writable classes.
Ensures backwards compatibility with code generation from cpp11 0.1.0 (#88)
push_back() now works more consistently with named arguments (#86)
cpp_source(), cpp_function() and cpp_eval() now support [[cpp11::linking_to()]] syntax to link to third party packages with C++ headers. (#48)
as_cpp<int>() and as_cpp<double>() now implicitly coerce between all 3 types of single NA values (#53).
list::const_iterator::operator*() added so iterators could be used on list objects (#60, @romainfrancois)
safe[] can now work with functions that return any type (#70, @bkietz)
The END_CPP macro now includes a catch(...) block to catch all C++ exceptions that do not inherit from std::exception (#47).
Improve consistency of inserting NA values in r_string objects (#45)
Added a NEWS.md file to track changes to the package.