diff options
author | William S Fulton <wsf@fultondesigns.co.uk> | 2015-12-31 18:04:16 +0000 |
---|---|---|
committer | William S Fulton <wsf@fultondesigns.co.uk> | 2015-12-31 18:04:16 +0000 |
commit | a8cf1eddf8007eb6eb9080a77289cdc5c6873e6c (patch) | |
tree | 468be1bbe0de1e5a641a338916a3e05fed6e48a0 /CHANGES | |
parent | 719c7d532ccf8489fb218ed1337ecfd9c8e15f33 (diff) | |
download | swig-a8cf1eddf8007eb6eb9080a77289cdc5c6873e6c.tar.gz |
Bump version to 3.0.9
Diffstat (limited to 'CHANGES')
-rw-r--r-- | CHANGES | 158 |
1 files changed, 158 insertions, 0 deletions
@@ -3,6 +3,164 @@ SWIG (Simplified Wrapper and Interface Generator) See the CHANGES.current file for changes in the current version. See the RELEASENOTES file for a summary of changes in each release. +Version 3.0.8 (31 Dec 2015) +=========================== + +2015-12-30: wsfulton + The pdf documentation is now generated by wkhtmltopdf and has colour + for the code snippets just like the html documentation! + +2015-12-23: ahnolds + [Python] Fixes for conversion of signed and unsigned integer types: + + No longer check for PyInt objects in Python3. Because PyInt_Check + and friends are #defined to the corresponding PyLong methods, this + had caused errors in Python3 where values greater than what could be + stored in a long were incorrectly interpreted as the value -1 with + the Python error indicator set to OverflowError. This applies to + both the conversions PyLong->long and PyLong->double. + + Conversion from PyLong to long, unsigned long, long long, and + unsigned long long now raise OverflowError instead of TypeError in + both Python2 and Python3 for PyLong values outside the range + expressible by the corresponding C type. This matches the existing + behavior for other integral types (signed and unsigned ints, shorts, + and chars), as well as the conversion for PyInt to all numeric + types. This also indirectly applies to the size_t and ptrdiff_t + types, which depend on the conversions for unsigned long and long. + +2015-12-19: wsfulton + [Python] Python 2 Unicode UTF-8 strings can be used as inputs to char * or + std::string types if the generated C/C++ code has SWIG_PYTHON_2_UNICODE defined. + +2015-12-17: wsfulton + Issues #286, #128 + Remove ccache-swig.1 man page - please use the CCache.html docs instead. + The yodl2man and yodl2html tools are no longer used and so SWIG no + longer has a dependency on these packages which were required when + building from git. + +2015-12-16: zturner/coleb + [Python] Fix Python3.5 interpreter assertions when objects are being + deleted due to an existing exception. Most notably in generators + which terminate using a StopIteration exception. Fixes #559 #560 #573. + If a further exception is raised during an object destruction, + PyErr_WriteUnraisable is used on this second exception and the + original exception bubbles through. + +2015-12-14: ahnolds/wsfulton + [Python] Add in missing initializers for tp_finalize, + nb_matrix_multiply, nb_inplace_matrix_multiply, ht_qualname + ht_cached_keys and tp_prev. + +2015-12-12: wsfulton + Fix STL wrappers to not generate <: digraphs. + For example std::vector<::X::Y> was sometimes generated, now + corrected to std::vector< ::X::Y >. + +2015-11-25: wsfulton + [Ruby] STL ranges and slices fixes. + + Ruby STL container setting slices fixes: + + Setting an STL container wrapper slice better matches the way Ruby + arrays work. The behaviour is now the same as Ruby arrays. The only + exception is the default value used when expanding a container + cannot be nil as this is not a valid type/value for C++ container + elements. + + Obtaining a Ruby STL container ranges and slices fixes: + + Access via ranges and slices now behave identically to Ruby arrays. + The fixes are mostly for out of range indices and lengths. + - Zero length slice requests return an empty container instead of nil. + - Slices which request a length greater than the size of the container + no longer chop off the last element. + - Ranges which used to return nil now return an empty array when the + the start element is a valid index. + + Ruby STL container negative indexing support improved. + + Using negative indexes to set values works the same as Ruby arrays, eg + + %template(IntVector) std::vector<int>; + + iv = IntVector.new([1,2,3,4]) + iv[-4] = 9 # => [1,2,3,9] + iv[-5] = 9 # => IndexError + +2015-11-21: wsfulton + [Ruby, Python] Add std::array container wrappers. + + These work much like any of the other STL containers except Python/Ruby slicing + is somewhat limited because the array is a fixed size. Only slices of + the full size are supported. + +2015-10-10: wsfulton + [Python] #539 - Support Python 3.5 and -builtin. PyAsyncMethods is a new + member in PyHeapTypeObject. + +2015-10-06: ianlancetaylor + [Go] Don't emit a constructor function for a director + class with an abstract method, since the function will + always panic. + +2015-10-01: wsfulton + Fix %shared_ptr support for private and protected inheritance. + - Remove unnecessary Warning 520: Derived class 'Derived' of 'Base' + is not similarly marked as a smart pointer + - Do not generate code that attempts to cast up the inheritance chain in the + type system runtime in such cases as it doesn't compile and can't be used. + Remove unnecessary warning 520 for %shared_ptr when the base class is ignored. + +2015-10-01: vkalinin + Fix #508: Fix segfault parsing anonymous typedef nested classes. + +2015-09-26: wsfulton + [Ruby] Add shared_ptr support + +2015-09-13: kkaempf + [Ruby] Resolve tracking bug - issue #225. + The bug is that the tracking code uses a ruby hash and thus may + allocate objects (Bignum) while running the GC. This was tolerated in + 1.8 but is invalid (raises an exception) in 1.9. + The patch uses a C hash (also used by ruby) instead. + +2015-09-09: lyze + [CFFI] Extend the "export" feature in the CFFI module to support + exporting to a specified package. + +2015-09-04: olly + [Python] Fix docstrings for %callback functions. + +2015-09-03: demi-rluddy + [Go] Removed golang stringing for signed/unsigned char + + Changed default handling of signed char* and unsigned char* to be + opaque pointers rather than strings, similarly to how other + languages work. + + Any existing code relying on treating signed char* or unsigned + char* as a string can restore the old behavior with typemaps.i by + using %apply to copy the [unchanged] char* behavior. + + *** POTENTIAL INCOMPATIBILITY *** + +2015-08-07: talby + [Perl] tidy -Wtautological-constant-out-of-range-compare warnings when building generated code under clang + +2015-08-07: xantares + [Python] pep257 & numpydoc conforming docstrings: + - Mono-line module docsstring + - Rewrite autodoc parameters section in numpydoc style: + https://github.com/numpy/numpy/blob/master/doc/HOWTO_DOCUMENT.rst.txt + - One line summary should end with "." + - Adds a blank line after class docstring + +2015-08-05: vadz + [Java] Make (char* STRING, size_t LENGTH) typemaps usable for + strings of other types, e.g. "unsigned char*". + Version 3.0.7 (3 Aug 2015) ========================== |