2536 Commits

Author SHA1 Message Date
Takatoshi Kondo
4a292193f2 Moved include byteswap.h to sysdep.h.
When compiling on Mac, byteswap.h is not included.
2014-10-12 01:22:10 +09:00
Takatoshi Kondo
1e7fbc0349 Removed wrong tr1:: from unordered containers. 2014-10-12 00:20:54 +09:00
Takatoshi Kondo
479d50bd93 Replaced uint64_t with apropriate types. 2014-10-11 23:45:21 +09:00
Nobuyuki Kubota
982787865f Merge pull request #127 from redboltz/clear_set_map
Fixed https://github.com/msgpack/msgpack-c/issues/76
2014-10-06 15:12:20 -07:00
Nobuyuki Kubota
bf4a57da34 Merge branch 'object-with-zone-copy' 2014-10-06 14:07:06 -07:00
Nobuyuki Kubota
2fcbe6116d Replace C-style casts with C++-style ones 2014-10-06 14:06:21 -07:00
Takatoshi Kondo
4c00f448aa Added a minimal forward declaration header file for pack/unpack overloading. 2014-10-01 14:33:02 +09:00
Takatoshi Kondo
3ddeb08e6e Added API versioning. 2014-10-12 23:21:25 +09:00
Takatoshi Kondo
0176dd760a Merge pull request #131 from shafik/master
Fixing undefined behavior introduced by the incorrect use of comma
2014-09-20 17:39:44 +09:00
Jakob Petsovits
d30548af3f Add support for copying an object to an object::with_zone.
With test.
2014-09-20 00:13:40 -04:00
Jakob Petsovits
ab12b2b2b9 Specialize operator<<(packer, T) also for object::with_zone.
It already existed for T = object, but if trying to pass an
object::with_zone it would match the generic operator<<() instead.
That operator calls o.msgpack_pack(), which doesn't exist and
causes a compile error.

(Re-)Using the overload for object is a better way to handle this.
2014-09-19 22:48:24 -04:00
Takatoshi Kondo
7ce69a362f Merge pull request #130 from jpetso/master
Support std::unordered_map/set even if using libstdc++ from GCC.
2014-09-20 00:06:51 +09:00
Takatoshi Kondo
cc08146b1c Merge pull request #132 from redboltz/support_travis_ci
Added travis-ci support.
2014-09-19 23:57:40 +09:00
Takatoshi Kondo
6896cc602f Added travis-ci support. 2014-09-19 23:42:07 +09:00
Shafik Yaghmour
d8f366daf2 Fixing undefined behavior introduced by the incorrect use of comma operator with the conditional operator. The middle expression in a conditional operator between the ? and : is implicitly parenthesized but the end expression is not. Since the comma operator as he lowest precendence( http://en.cppreference.com/w/cpp/language/operator_precedence) this means the conditional operator will be evaluated first and then the expression on the right hand side of the comma operator will be evaluated. This leads to undefined behavior because the last member of the union being updated will not be the member that will be used next which is strictly undefined in C++ although gcc and clang aloow this type punning as an extension but is clearly not portable behavior nor was this the intended behavior. instead of parenthesising the end expression I choose to use an if/else which is not subject to such easy to miss precdence issues. This Coliru live code demonstrates the bug with simple example: http://coliru.stacked-crooked.com/a/1041aaa8380feeaa the code also demonstrates the using the right warning flags gcc will generate a warning for this code. 2014-09-18 14:29:11 -04:00
Jakob Petsovits
ef649009e4 Support std::unordered_map/set even if using libstdc++ from GCC.
When passing the appropriate compiler flags, libstdc++ has
supported std::unordered_map and std::unordered_set for a
long time, even without the tr1 namespace. There's no reason to
limit support of non-tr1 containers to libc++ (clang) and MSVC.
2014-09-17 16:52:33 -04:00
Nobuyuki Kubota
cd03ab62f8 Merge pull request #128 from redboltz/cpp11_unpacked_return
Added additional unpack() APIs to support C++11 style programming.
2014-09-16 10:06:31 -07:00
Nobuyuki Kubota
44e41efa0c Merge pull request #126 from redboltz/add_zero_size_tuple_support
Added zero size tuple support.
2014-09-16 09:23:31 -07:00
Takatoshi Kondo
ecf4b09acd Merge pull request #129 from tarruda/fix-ext-8-16-32
Fix size packing/unpacking for EXT 8/16/32
2014-09-16 08:52:48 +09:00
Thiago de Arruda
d6122b4a18 Fix size packing/unpacking for EXT 8/16/32
For EXT 8/16/32, the "size" field was being incremented by 1 to account for the
type field, but according to the specification the size should only consider the
length of the data field.
2014-09-15 15:37:02 -03:00
Takatoshi Kondo
4fcb4777f3 Added additional unpack() APIs to support C++11 style programming. 2014-09-15 22:33:23 +09:00
Takatoshi Kondo
03ed30ce03 Fixed https://github.com/msgpack/msgpack-c/issues/76
Clear deserialized target containers.
As the result of this fix, all containers' deserialized behaviors become consistent.
2014-09-15 18:16:37 +09:00
Takatoshi Kondo
6e810df2a4 Added zero size tuple support. 2014-09-14 20:50:57 +09:00
Takatoshi Kondo
847a7852e5 Merge branch 'jpetso-master' 2014-09-10 10:59:49 +09:00
Jakob Petsovits
e265beef91 Install version.h even if doing an out-of-source build. 2014-09-09 16:21:38 -04:00
Takatoshi Kondo
0335df55e1 Merge pull request #123 from redboltz/add_memory_check
Add memory check
2014-09-09 14:53:23 +09:00
Takatoshi Kondo
dd4043a616 Removed unused variables. 2014-09-09 14:29:10 +09:00
Takatoshi Kondo
6a8412d6de Separated msgpack_test.cpp.
Renamed test files.
Decreased test's loop counter to avoid travis-ci no output timeout.
The number of loops is not so meaningful.
2014-09-09 14:15:21 +09:00
Takatoshi Kondo
7491c1996a Added memory checking for travis-ci cmake build. 2014-09-09 13:38:07 +09:00
Takatoshi Kondo
0ab14b4138 Merge pull request #121 from redboltz/fix_cmake_binary_dir
Fixed the path of the msgpack.pc to support structured build.
2014-09-09 10:27:52 +09:00
Takatoshi Kondo
6e77c6df11 Merge pull request #122 from redboltz/add_install_test
Added 'make install' as a test target.
2014-09-09 10:14:24 +09:00
Takatoshi Kondo
bcc91d78e9 Merge branch 'tarruda-ext-support-for-c' into poc/0.6 2014-09-09 10:09:02 +09:00
Takatoshi Kondo
32b65dba47 Added 'make install' as a test target. 2014-09-09 07:37:29 +09:00
Takatoshi Kondo
aa75e6ecd3 Fixed the path of the msgpack.pc to support structured build. 2014-09-09 06:28:52 +09:00
Takatoshi Kondo
dc679a2572 Merge pull request #119 from davidchappelle/master
Amendment to cmake changes for out of source builds
2014-09-09 06:21:38 +09:00
David Chappelle
b6bc7f7ecb Merge remote-tracking branch 'upstream/master'
Conflicts:
	CMakeLists.txt
2014-09-08 15:57:43 -04:00
Thiago de Arruda
896dead7cc Adapt msgpack_test.cpp EXT cases to msgpackc_test.cpp 2014-09-08 12:11:39 -03:00
Thiago de Arruda
dfa277a9e0 Add EXT support to C library 2014-09-08 12:11:38 -03:00
Jason Newton
501260eb54 correct paths for x86_64/generic installations 2014-09-08 22:27:38 +09:00
Takatoshi Kondo
370a9251f5 Merge pull request #117 from redboltz/removed_unused_member_variable
Removed unused member variable.
2014-09-08 17:27:16 +09:00
Takatoshi Kondo
7737f1cb77 Removed unused member variable. 2014-09-08 17:14:56 +09:00
Takatoshi Kondo
f72022037c Merge pull request #116 from redboltz/limited_travis_target
Limited branches for travis-ci.
2014-09-08 17:08:16 +09:00
Takatoshi Kondo
865528f778 Limited branches for travis-ci. 2014-09-08 17:01:04 +09:00
Takatoshi Kondo
d54048eabe Merge pull request #114 from redboltz/poc/0.6
Added travis-ci support.
2014-09-08 16:36:12 +09:00
Takatoshi Kondo
103a6c56e4 Added travis-ci support. 2014-09-08 16:38:44 +09:00
Takatoshi Kondo
11f1d5fbbd Fixed https://github.com/msgpack/msgpack-c/pull/113
When the original PR was applied, the resource leak had been fixed, but a dual free problem had been happened.

When msgpack_unpacker_next returns MSGPACK_UNPACK_CONTINUE, msgpack_unpacked::zone is not replaced. Then mespack_zone_free is called twice with the same object. msgpack_unpacked_destroy frees msgpack::zone when it is not NULL and set it to NULL.

Also, fixed memory leak (msgpack_sbuffer) on the test code.
2014-09-07 13:56:07 +09:00
Takatoshi Kondo
7ebdb63131 Merge pull request #113 from tarruda/fix-memory-leak
Fix memory leak in msgpack_unpacker_next
2014-09-07 12:45:23 +09:00
Thiago de Arruda
a186725213 Fix memory leak in msgpack_unpacker_next 2014-09-06 20:01:52 -03:00
David Chappelle
8615358515 Fixed out of source cmake builds to work correctly.
The out of source cmake build was not working correctly. In particular,
the main CMakeLists.txt was not installing from the correct location.
In the case of msgpack.pc, it was trying to install from the top repo
directory instead of from the cmake build directory. So you can now
build as follows:

    $ cd msgpack-c
    $ mkdir build
    $ cd build
    $ cmake ../
    $ make
    $ make install
2014-09-05 19:22:04 +00:00
Takatoshi Kondo
86ad026f11 Renamed the last digit of version string from RELEASE to REVISION.
Added REVISION macro and getting API.
2014-09-03 13:26:34 +09:00