天堂久久天堂AV色综合,久久久www免费人成精品,久久性生大片免费观看性

How are dtLinks created in NavMesh

金庆 — Fri, 18 Nov 2022 02:03:00 GMT

How are dtLinks created in NavMesh

(Jin Qing's Column, Nov., 2022)

`dtLink`

dtLink defines a link between 2 polygons. It is an internal data structure.

A dtLink is owned by its source polygon.

How link is used

The links are used in findPath(), to search all the neighbor polygons, from the current best polygon.

fields of `dtLink`

dtLink has 2 required fields:

neighbor: the neighbor polygon reference that is linked to.
edge: index of the polygon edge that owns this link. It is used in getPortalPoints() and raycast().

If the link is a boundary link, which links to other tile, then it has more fields:

side: defines on which side the link is. 8 sides. It is used in findPath() and raycast().
bmin: defines the minimum sub-edge area.
bmax: defines the maximum sub-edge area. bmin and bmax are used in portal finding.

How are links created

Links are not saved to navmesh file. They are created on navmesh loading in addTile().

dtNavMesh::addTile() add a tile to the navmesh. A tile is a square block of the map.

The main job of dtNavMesh::addTile() is to create links inside this tile and between this tile and other tiles. Actually a block indexed by (x, y) of constant size has many layers for multi-floor map. A dtMeshTile is actually a layer with an index of (x, y, layer).

5 steps to create all links:

1. Connect internal links

connectIntLinks() iterates all the polygons in the tile and create links for them by the help of neighbors information of the polygons.

2. Base off-mesh connections

baseOffMeshLinks() bases off-mesh connections to their starting polygons and connect connections inside the tile.

Off-mesh connection is between 2 points which are inside a tile or between 2 adjacent tiles.

For each off-mesh connection, there creates a specific polygon consisting of 2 vertices. See DT_POLYTYPE_OFFMESH_CONNECTION.

baseOffMeshLinks() creates 2 links for each off-mesh connection:

From the off-mesh connection polygon to the source polygon
From the source polygon to the off-mesh connection polygon

The destinaton polygon of the off-mesh connection is skipped here.

3. Connect external off-mesn links within this tile

connectExtOffMeshLinks() with the source and target tile be this tile.

connectExtOffMeshLinks() searches off-mesh connections that are from this tile to the target tile by checking the side direction of the connection.

It creates a link from off-mesh connection polygon to the target tile. For bidirectional connection, it also creates a link from the target polygon to the off-mesh connection polygon.

So for each off-mesh connection of this tile, 3 or 4 links are created by baseOffMeshLinks() and connectExtOffMeshLinks(), one on the source polygon, one on the destination polygon and 1/2 on the off-mesh connection polygon.

4. Connect with layers in current tile

For each layer other than this layer in this tile:

Create links from this layer to other layer
Create links from other layer to this layer
Create links of the off-mesh connection from this layer to other layer
Create links of the off-mesh connection from other layer to this layer

5. Connect with neighbour tiles

Check 9 neighbor tiles' layers, and create links just like previous step:

Create links from this layer to neighnor layer
Create links from neighnor layer to this layer
Create links of the off-mesh connection from this layer to neighnor layer
Create links of the off-mesh connection from neighnor layer to this layer

By now, for every off-mesh connection of all the loaded tiles, 3/4 links are created.

Reference

https://blog.csdn.net/icebergliu1234/article/details/80322392

金庆 2022-11-18 10:03 发表评论

C++ parameter passing rules

金庆 — Sat, 29 Oct 2022 03:01:00 GMT

C++ parameter passing rules

From: https://www.modernescpp.com/index.php/c-core-guidelines-how-to-pass-function-parameters

Parameter passing expression rules:

F.15: Prefer simple and conventional ways of passing information
F.16: For “in” parameters, pass cheaply-copied types by value and others by reference to const
F.17: For “in-out” parameters, pass by reference to non-const
F.18: For “consume” parameters, pass by X&& and std::move the parameter
F.19: For “forward” parameters, pass by TP&& and only std::forward the parameter
F.20: For “out” output values, prefer return values to output parameters
F.21: To return multiple “out” values, prefer returning a tuple or struct
F.60: Prefer T* over T& when “no argument” is a valid option

金庆 2022-10-29 11:01 发表评论

Naming Conventions for Accessors

金庆 — Thu, 22 Sep 2022 08:02:00 GMT

Naming Conventions for Accessors

from: C++ Programming Guidelines (cginternals.github.io)

When naming accessors within classes, non-trivial getters and queries, i.e., those that perform calculations, you should prepended get. All other getters have no prefix and setters have the set prefix.

class Object
{
public:
  int radius() const;
  void setRadius(int value);

  int getPerimeter();

private:
  int m_radius;
};

int Object::radius() const
{
    return m_radius;
}

void Object::setRadius(int value)
{
    m_radius = value;
}

int Object::getPerimeter()
{
    return 2 * pi * m_radius;
}

金庆 2022-09-22 16:02 发表评论

Visual Studio 2019 Compiler Hangs

金庆 — Sat, 31 Jul 2021 07:16:00 GMT

# Visual Studio 2019 Compiler Hangs

(金庆的专�?2021.7)

Discovered by my colleague Shen Yichai:
```
Share a interesting MS Build bug:
For file a.cpp, enable optimization for Win64 or XSX.
The MSBuild always compiling without error and ending (infinite compile).
The PS5 (clang) does not have this issue.
```

The simplified code:
```
struct Tag
{
   int v;
};

void Goo(Tag* r, Tag* a)
{
   r->v = a->v;
}

void Foo(Tag* pos0, Tag* pos)
{
   for (int j = 0; j < 4; j++)
   {
        if (j == 0) {
            for (int i = 0; i < 8; i++) {
                Goo(pos0++, pos);
                Goo(pos0++, pos);
            }
        }
        else {
            for (int i = 0; i < 8; i++) {
                Goo(pos0++, pos);
                Goo(pos0++, pos);
            }
        }
   }
}

int main()
{
   Foo(nullptr, nullptr);
}
```

The default release configuration can build correctly.
But if "Properties -> C/C++ -> Optimization -> Whole Program optimization" (/GL) of file or project is changed to "No",
the compilation will take a long time as 5 minutes.

```
Microsoft Visual Studio Community 2019
Version 16.9.5
VisualStudio.16.Release/16.9.5+31229.75
Microsoft .NET Framework
Version 4.8.04084
```

金庆 2021-07-31 15:16 发表评论

Fbx File Format Identifier

金庆 — Sun, 30 May 2021 01:55:00 GMT

# Fbx File Format Identifier

(金庆的专�?2021.5)

Print the list of FBX file format identifiers.

FBX SDK FbxExporter can set the file format identifier which is an int parameter.
The commnet says: User does not need to specify it by default.
If not specified, plugin will detect the file format according to file suffix automatically.

FBX SDK 2020 automatically set it to the latest binary format, which can not be parsed by "Autodesk FBX Converter 2013".
We need to specify the format identifier for FBX Converter 2013.
But there is no specification of this parameter.
We use the following code to display all the valid format identifiers and their descriptions.

```
#include
#include

int main()
{
    FbxManager* myManager = FbxManager::Create();
    assert(myManager);

    FbxIOSettings* ios = FbxIOSettings::Create(myManager, IOSROOT);
    myManager->SetIOSettings(ios);

    int lFormatCount = myManager->GetIOPluginRegistry()->GetWriterFormatCount();
    for (int lFormatIndex = 0; lFormatIndex < lFormatCount; lFormatIndex++)
    {
        FbxString lDesc = myManager->GetIOPluginRegistry()->GetWriterFormatDescription(lFormatIndex);
        printf("Format index %d: %s\n", lFormatIndex, (const char*)lDesc);
    }
    return 0;
}
```

Output:
```
Format index 0: FBX binary (*.fbx)
Format index 1: FBX ascii (*.fbx)
Format index 2: FBX encrypted (*.fbx)
Format index 3: FBX 6.0 binary (*.fbx)
Format index 4: FBX 6.0 ascii (*.fbx)
Format index 5: FBX 6.0 encrypted (*.fbx)
Format index 6: AutoCAD DXF (*.dxf)
Format index 7: Alias OBJ (*.obj)
Format index 8: Collada DAE (*.dae)
Format index 9: Biovision BVH (*.bvh)
Format index 10: Motion Analysis HTR (*.htr)
Format index 11: Motion Analysis TRC (*.trc)
Format index 12: Acclaim ASF (*.asf)
Format index 13: Acclaim AMC (*.amc)
Format index 14: Vicon C3D (*.c3d)
Format index 15: Adaptive Optics AOA (*.aoa)
Format index 16: Superfluo MCD (*.mcd)
```

FBX Converter 2013 will error "File is corrupted" for files exported as *.fbx file format: -1, 0, 2, 5.
Ascii format (1, 4) and FBX 6.0 binary (3) are OK.

金庆 2021-05-30 09:55 发表评论

查找内存错误

金庆 — Mon, 16 Dec 2019 10:03:00 GMT

查找内存错误

(金庆的专�?2019.12)

服务器进�E�有个偶发的崩溃�Q�breakpad 上传�?minidump 昄��调用栈不是出错的代码点，怀疑是内存错误�?br />
从日志分析可能触发出错的指��o�Q�然后在内网调试环境下测试，很幸�q�地�q�箋有几�ơ很�Ҏ��重现了�?br />
后来反复��试可知�Q�发送某个特�D?GM 指��o 200 �ơ就会出��C��ơ错误，一般会�?100 �ơ内出现�?br />但是以代码反复执行千�ơ万�ơ都不能重现�Q�估计与旉��有关�Q�需要手动操作�g�l�一�D�|��间后才会触发�?br />
出错点多数在 `boost::property_tree::ini_parser::read_ini()` 中，如下�Q?br />
```
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
Core was generated by `./GMServer.dbg'.
Program terminated with signal 11, Segmentation fault.
#0 0x00000000010ca227 in tcmalloc::SLL_Next(void*) ()
Missing separate debuginfos, use: debuginfo-install glibc-2.17-105.el7.x86_64 libgcc-4.8.5-4.el7.x86_64 libstdc++-4.8.5-4.el7.x86_64 zlib-1.2.7-15.el7.x86_64
(gdb) bt
#0 0x00000000010ca227 in tcmalloc::SLL_Next(void*) ()
#1 0x00000000010ca2b8 in tcmalloc::SLL_TryPop(void**, void**) ()
#2 0x00000000010ca715 in tcmalloc::ThreadCache::FreeList::TryPop(void**) ()
#3 0x00000000011ebe6c in tc_newarray ()
#4 0x00007efddb7d4c69 in std::string::_Rep::_S_create(unsigned long, unsigned long, std::allocator const&) () from /lib64/libstdc++.so.6
#5 0x0000000000a18153 in std::string::_S_construct (__beg=0x3f6c8b9 "LobbyServer]", __end=0x3f6c8c4 "]", __a=...) at /usr/include/c++/4.8.2/bits/basic_string.tcc:138
#6 0x00007efddb7d641c in std::basic_string, std::allocator >::basic_string(std::string const&, unsigned long, unsigned long) ()
   from /lib64/libstdc++.so.6
#7 0x00007efddb7d6462 in std::string::substr(unsigned long, unsigned long) const () from /lib64/libstdc++.so.6
#8 0x0000000000a3c3be in boost::property_tree::ini_parser::read_ini > > (stream=..., pt=...)
    at /var/tmp/src/f4f4f712-7894-4d98-83dd-b91be8e0555e/Linux-Debug/003_servers/../000_BaseLib/3RdParty/boost/include/boost/property_tree/ini_parser.hpp:111
#9 0x0000000000a3b2f0 in boost::property_tree::ini_parser::read_ini > > (
    filehelp="cfg.ini", pt=..., loc=...)
    at /var/tmp/src/f4f4f712-7894-4d98-83dd-b91be8e0555e/Linux-Debug/003_servers/../000_BaseLib/3RdParty/boost/include/boost/property_tree/ini_parser.hpp:169
...
#27 0x00007efddba27dc5 in start_thread () from /lib64/libpthread.so.0
#28 0x00007efddaf341cd in clone () from /lib64/libc.so.6
(gdb) q
```

但是 property_tree 是实践证明可靠的库，暂时不去怀疑它�?br />其他代码�?uWS 处理HTTP ��h��Q�用 hiredis ��d�� Redis�?br />�q�些库先假定是正��的�Q�先查看自己的代码，因�ؓ相关代码较少�Q�可以肉眼查错，可惜只查��C��无关的小错误�?br />
然后到处��d��调试日志�Q�跟�t�变量的构造与析构�?br />因�ؓ uWS 中用��C��一个用戯��定义数据�Q�有一�?new/delete 操作�Q�容易出内存错误�Q�重点就是这个数据是否存在野指针�?br />但是看上��L��常�?br />
肉眼查代码不行，��p��用工具了�?br />
首先��d�� -fstack-protector-all �~�译参数�Q�这个参数其实无论如何都应该早就加上的，用来��函数调用时栈破坏�?br />同时在自定义用户数据使用和析构时��查自�w�数据是否有效，如某个成员变量��L��设�ؓ 0x5a, 析构时才把它�|��ؓ 0xcc.
崩溃依旧�Q�但上述内存��查正常�?br />
再��用一�?Address Sanitizer, 只需要添加编译选项 -fsanitize=address 卛_��?br />它用来检查堆内存是否存在非法操作�?br />但是�q�个不能�?tcmalloc ��q��Q�我需要更改现有代码，然后链接 asan 库，�l�于可以�q�行��h��了�?br />如果存在 tcmalloc 调用�Q�进�E�启动时��׃��崩溃�?br />
效果很好�Q�准��报�?"double-free" 错误�Q?br />```
[jinqing@host-192-168-21-31 bin]$ gdb GMServer.dbg
GNU gdb (GDB) Red Hat Enterprise Linux 7.6.1-80.el7
Copyright (C) 2013 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law. Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-redhat-linux-gnu".
For bug reporting instructions, please see:
...
Reading symbols from /home/jinqing/valkyrie/Runtime_ZT_QA/bin/GMServer.dbg...done.
(gdb) r
Starting program: /home/jinqing/valkyrie/Runtime_ZT_QA/bin/GMServer.dbg
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
iLogLevel:1
[New Thread 0x7ffff3724700 (LWP 1010)]
[New Thread 0x7ffff2c1d700 (LWP 1011)]
[New Thread 0x7ffff2116700 (LWP 1012)]
[New Thread 0x7ffff160f700 (LWP 1013)]
[New Thread 0x7ffff0b08700 (LWP 1014)]
=================================================================
==1005== ERROR: AddressSanitizer: attempting double-free on 0x6004000270d0:
    #0 0x7ffff4e60dd9 (/usr/lib64/libasan.so.0.0.0+0x15dd9)
    #1 0x1570e8d (/home/jinqing/valkyrie/Runtime_ZT_QA/bin/GMServer.dbg+0x1570e8                                                                                                         d)
0x6004000270d0 is located 0 bytes inside of 5-byte region [0x6004000270d0,0x6004                                                                                                         000270d5)
freed by thread T0 here:
    #0 0x7ffff4e60dd9 (/usr/lib64/libasan.so.0.0.0+0x15dd9)
    #1 0x1570e8d (/home/jinqing/valkyrie/Runtime_ZT_QA/bin/GMServer.dbg+0x1570e8                                                                                                         d)
previously allocated by thread T0 here:
    #0 0x7ffff4e60ef9 (/usr/lib64/libasan.so.0.0.0+0x15ef9)
    #1 0x157179e (/home/jinqing/valkyrie/Runtime_ZT_QA/bin/GMServer.dbg+0x157179                                                                                                         e)
Thread T4 created by T0 here:
    #0 0x7ffff4e55a0a (/usr/lib64/libasan.so.0.0.0+0xaa0a)
    #1 0x13fc796 (/home/jinqing/valkyrie/Runtime_ZT_QA/bin/GMServer.dbg+0x13fc79                                                                                                         6)
==1005== ABORTING
[Thread 0x7ffff160f700 (LWP 1013) exited]
[Thread 0x7ffff2116700 (LWP 1012) exited]
[Thread 0x7ffff2c1d700 (LWP 1011) exited]
[Thread 0x7ffff3724700 (LWP 1010) exited]
[Thread 0x7ffff7fe77c0 (LWP 1005) exited]
[Inferior 1 (process 1005) exited with code 01]
Missing separate debuginfos, use: debuginfo-install glibc-2.17-105.el7.x86_64 li                                                                                                         basan-4.8.5-39.el7.x86_64 libgcc-4.8.5-4.el7.x86_64 libstdc++-4.8.5-4.el7.x86_64                                                                                                          zlib-1.2.7-15.el7.x86_64
(gdb)
```

因�ؓ用了 gcc 4.8, 所以asan没有函数名输出，gcc 4.9 以上��׃��有函数名了�?br />但是不要紧，�?nm 可以搜到函数名：

```
[jinqing@host-192-168-21-31 bin]$ nm GMServer.dbg | grep 1570e
0000000001570e50 T freeReplyObject
0000000001570ef0 T redisReaderFree
[jinqing@host-192-168-21-31 bin]$ nm GMServer.dbg | grep 15717
0000000001571750 t createStringObject
[jinqing@host-192-168-21-31 bin]$ nm GMServer.dbg | grep 13fc7
00000000013fc700 t _ZN5boost6detail23set_current_thread_dataEPNS0_16thread_data_baseE
00000000013fc7d0 T _ZN5boost6thread21start_thread_noexceptERKNS_17thread_attributesE
00000000013fc750 T _ZN5boost6thread21start_thread_noexceptEv
00000000013fc740 T _ZN5boost6threadC1Ev
00000000013fc740 T _ZN5boost6threadC2Ev
[jinqing@host-192-168-21-31 bin]$ ^C
```

昄��?freeReplyObject() 调用�?"double-free".

有了�q�个提示�Q�查代码��q��单了。原来查 uWS 的用��h��?new/delete 是方向性错误，
Redis 也有�?delete 操作�Q�即�q�回�?ReplyObject 需要调�?freeReplyObject().
直接搜烦一遍就扑ֈ�了那个多余的 freeReplyObject.

原来�?ReplyObject 保存��Z��个成员变量，每次 Redis 操作完成后需要调�?freeReplyObject() �q�将该成员设为空�?br />同时析构函数中也会判断该成员�Q�非�I�则调用 freeReplyObject().
但是有个�?60s 执行一�ơ的 Ping 操作�Q�调用了 freeReplyObject(), 但是没有�|�空�?br />同时 GM 指��o也触发了一��?Redis 对象重置�Q�再�ơ调用了 freeReplyObject()�?br />
正确的做法是不用成员变量�Q�每�ơ都用��时变量就行了�Q�但是用成员变量可以��写一行声明，写代码较方便�?br />��Z��改代码�Q�仍然保持�ؓ成员变量�Q�仅把缺的置�I��上了�?br />同时析构中的 freeReplyObject() 是没必要的，�Ҏ��断言成员为空�?br />
��M��一行代码，然后��试�Q�发现错误不出现了。如果不加这行置�I�，��?Ping 间隔�~�小�Q�结果出错概率大大增加了�?br />由此��认�Q�错误点已找到�ƈ修正了�?br />

金庆 2019-12-16 18:03 发表评论

std::thread 中的异常会丢��p��用栈

金庆 — Thu, 26 Sep 2019 09:19:00 GMT

# std::thread 中的异常会丢��p��用栈

(金庆的专�?2019.9)

��d��C��的异常生成core文�g能看到调用栈�Q?br />但是 std::thread 子线�E�中的异常生�?core 文�g的调用栈如下�Q?br />
```
Program received signal SIGABRT, Aborted.
[Switching to Thread 0x7f18f00ab700 (LWP 32340)]
0x00007f19007335f7 in raise () from /lib64/libc.so.6
(gdb) bt
#0 0x00007f19007335f7 in raise () from /lib64/libc.so.6
#1 0x00007f1900734ce8 in abort () from /lib64/libc.so.6
#2 0x00007f19010379d5 in __gnu_cxx::__verbose_terminate_handler() () from /lib64/libstdc++.so.6
#3 0x00007f1901035946 in ?? () from /lib64/libstdc++.so.6
#4 0x00007f1901035973 in std::terminate() () from /lib64/libstdc++.so.6
#5 0x00007f190108c2b5 in ?? () from /lib64/libstdc++.so.6
#6 0x00007f19012e7dc5 in start_thread () from /lib64/libpthread.so.0
#7 0x00007f19007f41cd in clone () from /lib64/libc.so.6
(gdb)
```

调用栈丢�׃��调试无法扑ֈ�问题炏V�?br />
据说 GCC 8 修正了该问题�?br />
参考：
* [C++ uncaught exception in worker thread](https://stackoverflow.com/questions/48535100/c-uncaught-exception-in-worker-thread)
* [Bug 55917 - Impossible to find/debug unhandled exceptions in an std::thread](https://gcc.gnu.org/bugzilla/show_bug.cgi?id=55917)

金庆 2019-09-26 17:19 发表评论

用赋��g��?protobuf CopyFrom()

金庆 — Thu, 04 Apr 2019 09:57:00 GMT

# 用赋��g��?protobuf CopyFrom()

�C�Z��Q�[Replace protobuf CopyFrom with assignment](https://github.com/tensorflow/tensorflow/commit/9501c4104125fb8c2c2d2e837fc2dd8a24034d52)

protobuf 生成�?C++ 代码中，因�ؓ CopyFrom() 可以接受��M�� Message 作�ؓ参数�Q?br />所以有可能�?个不同类型的消息之间复制�?br />
```
void CopyFrom(const ::google::protobuf::Message& from) final;
void CopyFrom(const PlayerData& from);
```

而赋值操作可以保证类型正��?br />
```
class PlayerData : public ::google::protobuf::Message {
public:
...
inline PlayerData& operator=(const PlayerData& from) {
    CopyFrom(from);
    return *this;
}
#if LANG_CXX11
inline PlayerData& operator=(PlayerData&& from) noexcept {
    ...
}
#endif
```

�c�d��不一致时�~�译会报错：
```
error: no match for ‘operator=’ (operand types are ‘a::PlayerData’ and ‘a::HeroInfo’)
```

发现自己用了多年�?CopyFrom() 都是错误的��用�?img src ="http://www.shnenglu.com/jinq0123/aggbug/216338.html" width = "1" height = "1" />

金庆 2019-04-04 17:57 发表评论

vs2017 linux �~�译输出�Ҏ�� vs 格式

金庆 — Wed, 21 Nov 2018 02:57:00 GMT

vs2017 linux �~�译输出�Ҏ�� vs 格式

(金庆的专�?2018.11)

1 #!/usr/bin/python
2 # -*- coding: utf-8 -*-
3
4 '''
5 gcc2vs.py
6
7 功能�Q?br /> 8 ��剪切板中gcc的编译输出格式�{成vs格式�Q�用于vs跌��{到错误行�?br /> 9 vs2017 linux �~�译输出�?nbsp;gcc 格式�Q�无法点击蟩转，如：
10 /var/tmp/src/dbe/Linux-Debug/Src/Team.cpp:16:1: 错误�Q?#8216;x’不是一个类型名
11 ��{为vs格式, �?br />12 /var/tmp/src/dbe/Linux-Debug/Src/Team.cpp(16):1: 错误�Q?#8216;x’不是一个类型名
13
14 如何使用�Q?br />15
16 首先��d��?nbsp;python, �q�安�?nbsp;pyperclip
17 pip install pyperclip
18
19 假设本文件�ؓ d:/tools/gcc2vs.py,
20 vs讄��外部工具�Q�工�?>外部工具->��d��
21   标题�Q�gcc2vs(&V)
22   命��o�Q�python.exe
23   参数�Q�d:/tools/gcc2vs.py
24   选中"使用输出�H�口"
25
26 参考：VS2010手动��d��外部工具和快捷键
27 https://www.cnblogs.com/ChinaHook/p/4698733.html
28
29 当Linux构徏输出后，点击输出�H�口�Q�ctrl-A 选择全部�Q�ctrl-C 复制输出到剪切板�Q?br />30 然后 alt-T,V �q�行��d��的外部工�?nbsp;gcc2vs(&V), 更改输出格式�Q�然后就可以点击错误跌��{了�?br />31 '''
32
33 import re
34 import pyperclip
35
36 # 待替换的格式
37 pattern = re.compile(r'/var/tmp/src/..-.-.-.-/Linux-Debug/(.*):([0-9]*):([0-9]*): ')
38
39 test_lines_src = '''
40 /var/tmp/src/db71a8ec-90bb-2838-98df-2dd35e71166e/Linux-Debug/003_servers/103_LobbyServer/Src/Team.cpp:16:1: 错误�Q?#8216;x’不是一个类型名
41 生成��p�|�?br />42 '''
43 test_lines_dst = '''
44 003_servers/103_LobbyServer/Src/Team.cpp(16):1: 错误�Q?#8216;x’不是一个类型名
45 生成��p�|�?br />46 '''
47 assert test_lines_dst == re.sub(pattern, r'\1(\2):\3: ', test_lines_src)
48
49 # 剪切板中的gcc格式输出
50 src = pyperclip.paste()
51 # 转成vs格式
52 dst = re.sub(pattern, r'\1(\2):\3: ', src)
53 print(dst)
54

金庆 2018-11-21 10:57 发表评论

�?LiteIDE ��d��选中标记

金庆 — Sat, 06 Jan 2018 03:05:00 GMT

�?LiteIDE ��d��选中标记

(金庆的专�?2018.1)

"[Eclipse Mark Occurrences](http://www.eclipse.org/pdt/help/html/mark_occurrences.htm)"
可以在滚动条旁边昄��文中所有选中的标讎ͼ�
可以选中某个变量昄��有多��引用。不�q?GoClipse 没有 "Mark Occurrences" 功能�?br />VS code 有此功能�?br />
LiteIDE 对于build错误会显�C��U�标讎ͼ�所以感觉可以依此实�?Mark Occurrences"�?br />
首先扑ֈ�了打标记的功能。标记用到颜�Ԍ��所以搜'QColor', 扑ֈ�了：
```c++
QColor markTypeColor(LiteApi::EditorNaviagteType type)
```

相关的接口会用到�Q?br />```c++
    void insertNavigateMark(int line, LiteApi::EditorNaviagteType type, const QString &msg, const char* tag);
    void clearAllNavigateMark(LiteApi::EditorNaviagteType types, const char *tag);
```

��L��选中时触发标记的代码�Q�选中或搜索时会有圆角框圈出，�?find", 扑ֈ��Q?br />```c++
if (!m_findExpression.isEmpty()) {
    if (!findInBlock(block,m_findExpression,pos,m_findFlags,cur)) {
        break;
    }
    ...
    painter.drawRoundedRect(offsetX+left,r.top()+l.y(),right-left,l.height(),3,3);
} else if (!m_selectionExpression.isEmpty()) {
    if (!findInBlock(block,m_selectionExpression,pos,QTextDocument::FindWholeWords,cur)) {
        break;
    }
    ...
    painter.drawRoundedRect(offsetX+left,r.top()+l.y(),right-left,l.height(),3,3);
}
```

发现��L��框仅在可视区域。箋�l�查�?m_findExpression �?m_selectionExpression 更改处，
��d�� 'updateNavigateMarks()'

```c++
void LiteEditorWidgetBase::setFindOption(LiteApi::FindOption *opt)
{
    ...
+   updateNavigateMarks(LiteApi::EditorNavigateFind);
    viewport()->update();
}
```

```c++
void LiteEditorWidgetBase::slotSelectionChanged()
{
    ...
        m_selectionExpression.setPattern(pattern);
+       updateNavigateMarks(LiteApi::EditorNavigateSelection);
        viewport()->update();
    ...
}
```

��L��?个新的类型：
```c++
enum EditorNaviagteType{
    EditorNavigateNormal = 1,
    EditorNavigateWarning = 2,
    EditorNavigateError = 4,
    EditorNavigateReload = 8,
+   EditorNavigateFind = 16,
+   EditorNavigateSelection = 32,
    EditorNavigateBad = EditorNavigateWarning|EditorNavigateError
};
```

搜烦 `EditorNavigateWarning`, 扑ֈ�因新增类型须更改优先�U�表和颜色函数�?br />
```c++
const int PRIORITYLIST_LENGTH = 7;
const LiteApi::EditorNaviagteType MARKTYPE_PRIORITYLIST[PRIORITYLIST_LENGTH] = {
        ..., LiteApi::EditorNavigateFind, LiteApi::EditorNavigateSelection, ...
    };
...
inline QColor markTypeColor(LiteApi::EditorNaviagteType type) {
    switch(type) {
    ...
    case LiteApi::EditorNavigateNormal:
        return Qt::darkGreen;
    case LiteApi::EditorNavigateReload:
        return Qt::darkBlue;
    }
}
```

更新标记时先清空�Q�然后逐行搜烦��d��标记�Q?br />```c++
// Update selections or find marks.
void LiteEditorWidgetBase::updateNavigateMarks(LiteApi::EditorNaviagteType type)
{
    clearAllNavigateMark(type, "");
    ...

    QTextDocument *doc = this->document();
    for (QTextBlock it = doc->begin(); it != doc->end(); it = it.next())
    {
        if (!needToMarkBlock(it, type))
            continue;
        int lineNumber = it.blockNumber() + 1;
        insertNavigateMark(lineNumber, type, QString("%1: %2").arg(lineNumber).arg(it.text()), "");
    }
}

bool LiteEditorWidgetBase::needToMarkBlock(
    const QTextBlock &block, LiteApi::EditorNaviagteType type) const
{
    ...
    if (LiteApi::EditorNavigateFind == type)
        return findInBlock(block, m_findExpression, pos, m_findFlags, cur);
    if (LiteApi::EditorNavigateSelection == type)
        return findInBlock(block, m_selectionExpression, pos,
                           QTextDocument::FindWholeWords, cur);
    return false;
}
```

已合�q�主�qԌ��
```
Revision: 43f4954b0b802eccbbf451136be600bfcec71f27
Author: Jin Qing
Date: 18.1.5 19:08:26
Message:
Add "Mark Occurrences" function that marks selections and findings.

----
Modified: liteidex/src/api/liteeditorapi/liteeditorapi.h
Modified: liteidex/src/plugins/liteeditor/liteeditorwidgetbase.cpp
Modified: liteidex/src/plugins/liteeditor/liteeditorwidgetbase.h
```

金庆 2018-01-06 11:05 发表评论

asio 协程�?yield

金庆 — Thu, 07 Dec 2017 06:51:00 GMT

asio 协程�?yield

(金庆的专�?2017.12)

https://stackoverflow.com/questions/26127458/yielding-in-boost-asio-stackful-coroutine

Asio spawn() 可以产生一个协�E�，协程中可以调�?async_read(..., yield), async_write(..., yield), 但是不知道如何主动释放控制权(yield)?

asio::spawn(strand_, [this, self](asio::yield_context yield)

{

while (!computationFinished)

{

computeSomeMore();

yield; // WHAT SHOULD THIS LINE BE?

}

�{�案是：

iosvc.post(yield);

其他�q�可以是

iosvc.poll_one();

iosvc.poll();

应该�?post(yield) 最合适�?/p>

... polling the io_service avoids the context switch overhead, but unhandled exceptions from handlers will unwind and destroy the coroutine.

金庆 2017-12-07 14:51 发表评论

Lua和C++之间调用效率��试

金庆 — Wed, 30 Aug 2017 09:25:00 GMT

Lua和C++之间调用效率��试

(金庆的专�?2017.8)

仿照 http://www.cnblogs.com/archy_yu/p/3185608.html �?Lua �?C++ 调用�q�行��试�?br />
代码见：https://github.com/jinq0123/TimerLuaIntf

使用 LuaIntf �l�定 Lua �?C++。用 boost timer 计时�?br />依赖�?lua-cpp, lua-intf, boost-timer �?conan 安装�?br />conan 会下载源码，�~�译�Q�然后生�?conanbuildinfo.props �l?VS 导入�Q?br />其中讑֥�了所�?include, lib 目录�Q�链接库�Q�运行库�?br />
代码大概如下�Q?br />
   cout << "C++ calls lua add() many times:\n";
   {
       boost::timer::auto_cpu_timer t;
       for (int i = 0; i < COUNT; ++i)
           test.dispatchStatic("add", 123, 456);
   }
   cout << "C++ calls lua add_times() once:\n";
   {
       boost::timer::auto_cpu_timer t;
       test.dispatchStatic("add_times", 123, 456, COUNT);
   }
   cout << "Lua calls C++ add() many times:\n";
   {
       boost::timer::auto_cpu_timer t;
       test.dispatchStatic("test_c_add", 123, 456, COUNT);
   }
   cout << "Lua calls C++ add_times() once:\n";
   {
       boost::timer::auto_cpu_timer t;
       test.dispatchStatic("test_c_add_times", 123, 456, COUNT);
   }

��试4�U�调用：
* C++ 调用 1kw ��?lua add()
* C++ 调用 1 ��?lua add_times(), 其中调用 add() 1kw ��?br />* Lua 调用 C++ add() 1kw ��?br />* Lua 调用 C++ add_times() 1 �ơ，其中调用 add() 1kw ��?br />
输出如：
C++ calls lua add() many times:
2.759473s wall, 2.761218s user + 0.000000s system = 2.761218s CPU (100.1%)
C++ calls lua add_times() once:
0.436400s wall, 0.436803s user + 0.000000s system = 0.436803s CPU (100.1%)
Lua calls C++ add() many times:
0.535802s wall, 0.530403s user + 0.000000s system = 0.530403s CPU (99.0%)
Lua calls C++ add_times() once:
0.000005s wall, 0.000000s user + 0.000000s system = 0.000000s CPU (n/a%)

�l�论是：
* C++ 调用 Lua 可达 3百万��?s
* Lua 内部调用函数可达 2千万�ơ强/s
* Lua 调用 C++ 函数可达 2千万�ơ弱/s

金庆 2017-08-30 17:25 发表评论

conan-transit服上的库列表

金庆 — Sat, 05 Aug 2017 05:14:00 GMT

摘要: conan-transit服上的库列表因�ؓ获取列表比较慢，所以获取后在此记录�Q�以备查找。conan-transit 是个只读库，不会有更新。新的库��上传到 conan-center. conan是C/C++包管理器。conan search --remote conan-transitExisting package recipes:7z_installer/0.1@lasote/testingA... 阅读全文

金庆 2017-08-05 13:14 发表评论

金庆 — Mon, 31 Jul 2017 07:00:00 GMT

Premake 生成 Makefile 的缺省配�|?br />
(金庆的专�?2017.7)

premake5.exe --os=linux gmake

生成�?Makefile 中有�?config, �?make 生成release版本时��用以下命�?

make config=release_x64

在premake5.lua中可以如下设�|�，使config�~�省�?release_x64�Q?br />
workspace "AppWorkspace"
   configurations { "Release", "Debug" }
   platforms { "x64", "x32" }
   ...

即缺省配�|��ؓ configurations �?platforms 的首个选项�?br />
生成�?Makefile 中会如下初始�?config:

ifndef config
config=release_x64
endif

�q�样 make ��׃��~�省生成 release_x64�?/div>

金庆 2017-07-31 15:00 发表评论

VC6工程因行��格式无法�{换到VS2015

金庆 — Wed, 07 Jun 2017 02:22:00 GMT

VC6工程因行��格式无法�{换到VS2015

(金庆的专�?2017.6)

参考：
https://connect.microsoft.com/VisualStudio/feedback/details/546432/problems-with-vs2010rc-doing-conversion-of-vs6-dsp-file-vcproj-file

apr
消息
apr.dsp: 无法转换��目。请��保�q�是一个有效的 Visual C++ 6.0 ��目�?
apr.dsp: ��目升��p�|�?
apr.dsp: 转换��目文�g“E:\temp\apr-1.5.2\apr-1.5.2\apr.dsp”�?

原来下蝲的是 zip 包可以正常�{换升�U��?br />现在下蝲�?bz 压羃包，解开后是Unix格式�Q�所以出现无法�{换�?br />转成Windows格式后就可以了�?br />

金庆 2017-06-07 10:22 发表评论

std::hash实现太简单分布不匀

金庆 — Fri, 26 May 2017 04:00:00 GMT

std::hash实现太简单分布不匀

(金庆的专�?2017.5)

#include
#include

using namespace std;

int main()
{
    std::hash hasher;
    cout << hasher(2) << endl;
    cout << hasher(3) << endl;
    cout << hasher(4) << endl;

    return 0;
}

输出�?br />jinqing@server:~/test$ g++ main.cpp -std=c++11
jinqing@server:~/test$ ./a.out
2
3
4

查看实现�Q?usr/include/c++/5/bits/functional_hash.h

      operator()(_Tp __val) const noexcept              \
      { return static_cast(__val); }            \

所以对分布有要求的�Q�应该��用自��q��hash, 不要使用 std::hash.

boost::hash 的实��C��是简单取��|��
boost_1_60_0/boost/functional/hash/hash.hpp

    template
    typename boost::hash_detail::basic_numbers::type hash_value(T v)
    {
        return static_cast(v);
    }

Boost说明了hash用于STL容器�Q�而不是其它�?br />   This hash function is designed to be used in containers based on the STL and is not suitable as a general purpose hash function.

VS2015会��?FNV-1a

   size_t operator()(const argument_type& _Keyval) const
       {   // hash _Keyval to size_t value by pseudorandomizing transform
       return (_Hash_seq((const unsigned char *)_Keyval.c_str(),
           _Keyval.size() * sizeof (_Elem)));
       }

   inline size_t _Hash_seq(const unsigned char *_First, size_t _Count)
       {   // FNV-1a hash function for bytes in [_First, _First + _Count)
           ...

�?FNV-1a 也不是通用�?hash 函数�Q�如果输入值相�q�，则其输出��g��相近�?br />

金庆 2017-05-26 12:00 发表评论

用preload加蝲Lua导出模块

金庆 — Wed, 10 May 2017 08:11:00 GMT

用preload加蝲Lua导出模块

(金庆的专�?2017.5)

参考：
How to make exported module non-global?
https://github.com/SteveKChiu/lua-intf/issues/135

动态库可以�q�样导出模块�Q?br />
    extern "C"
    #if defined(_MSC_VER) || defined(__BORLANDC__) || defined(__CODEGEARC__)
    __declspec(dllexport)
    #endif
    int luaopen_modname(lua_State* L)
    {
        LuaRef mod = LuaRef::createTable(L);
        LuaBinding(mod)
            ...;
        mod.pushToStack();
        return 1;
    }

如果不是动态库�Q�可以这样导出全局模块 c_util�Q?br />
int main()
{
   ...
   LuaBinding(L).beginModule("c_util")
       .addFunction("foo", []() { return 123; })
   .endModule();
   ...
}

如果不想让它成�ؓ全局模块�Q�则需要在 package.preload 表中注册一个加载函�?

Lua�E�序设计 �W?�?英文�?programming in lua 3ed

The preload searcher allows the definition of an arbitrary function to load a module.
It uses a table, called package.preload, to map module names to loader functions.
When searching for a module name, this searcher simply looks for the given name in the table.
If it finds a function there, it returns this function as the module loader.
Otherwise, it returns nil.
This searcher provides a generic method to handle some non-conventional situations.
For instance, a C library statically linked to Lua can register its luaopen_ function into the preload table,
so that it will be called only when (and if) the user requires that module.
In this way, the program does not waste time opening the module if it is not used.

代码�C�Z��Q?br />
    extern "C" int open_my_module(lua_State* L)
    {
        LuaRef mod = LuaRef::createTable(L);
        LuaBinding(mod)
            .addFunction("get_my_svr_id", &Util::GetMySvrId)
            ;
        mod.pushToStack();
        return 1;
    }

   int main()
   {
       ...
       LuaRef table(L, "package.preload");
       table["c_util"] = LuaRef::createFunctionWith(L, open_my_module);
       ...
   }

Lua ��试�Q?br />   assert(c_util == nil)
   local t = require("c_util")
   assert("table" == type(t))
   assert("function" == type(t.get_my_svr_id))

金庆 2017-05-10 16:11 发表评论

金庆 — Sun, 07 May 2017 02:38:00 GMT

grpc++不支持异步多�ơ写�?/p>

(金庆的专�?2017.5)

见：

Multi ClientAsyncWriter::Write() calls will crash. #7659
https://github.com/grpc/grpc/issues/7659

why the continuous write to async streaming crash? #4007
https://github.com/grpc/grpc/issues/4007

write stream error:GRPC_CALL_ERROR_ALREADY_INVOKED
https://github.com/grpc/grpc/issues/3953

grpc_cb 支持真正的异步写�?br />https://github.com/jinq0123/grpc_cb

金庆 2017-05-07 10:38 发表评论

支持 proto3 �?lua �l�定�?LuaPbIntf

金庆 — Tue, 25 Apr 2017 03:43:00 GMT

支持 proto3 �?lua �l�定�?LuaPbIntf

(金庆的专�?2017.4)

protobuf的lua�l�定库之前一直用 pbc�Q�但是该库已�l�不更新了�?br />惌��d�� proto3 �?map , �q�有支持 service, 发现代码太高深，无法下手�?br />
pbc因�ؓ用c代码实现�?protobuf 的功能，所以代码量很大�?br />而对�?c++ 来说�Q�只需要利�?protobuf 库中动态消息就可以实现 lua 所需功能�?br />
luapb ��是直接�?protobuf 库封装后导出lua实现的。该库仅支持 proto2.

�?luapb 基础上实现对 proto3 的支持，��是 LuaPbIntf 库�?br />LuaPbIntf 库利用了 lua-intf 库来实现 c++ �?lua �l�定�Q�所以代码量�?luapb 又少了许多，
�q�且可读性大大增强，更改��d��新功能也��更方便了�?br />
�?pbc, lubpb 一��P��LuaPbIntf 也是动态加�?proto 文�g�Q�不需要代码生成�?br />
LuaPbIntf 采用 lua table 来表�C?pb 消息�Q�不�?pbc 中的代理表，而只是普通表�?br />
��目地址�Q?a >https://github.com/jinq0123/LuaPbIntf

�C�Z��Q?br />
local pb = require("luapbintf")

pb.import_proto_file("test.proto")

local msg = { uid = 12345 }
local sz = pb.encode("test.TestMsg", msg)

local msg2 = pb.decode("test.TestMsg", sz)
assert(msg2.uid == 12345)

proto3 map �C�Z��Q?br />
local msgs = {}
msgs["k1"] = {}
msgs["k2"] = {}
pb.encode("test.TestMsg", { msgs = msgs })

rpc service 支持:

assert(pb.get_rpc_input_name("test.Test", "Foo") == "test.TestMsg")
assert(pb.get_rpc_output_name("test.Test", "Foo") == "test.CommonMsg")

金庆 2017-04-25 11:43 发表评论

金庆 — Mon, 17 Apr 2017 06:40:00 GMT

��proto文�g按包名分子目�?br />
(金庆的专�?2017.4)

服务器客��L��之间的protobuf协议定义在客��L��与服务器公共目录下，包名为rpc.
服务器内部协议定义在服务器目录下�Q�包名�ؓsvr.
rpc.EmptyMsg �?svr.EmptyMsg 分别定义在各自的根目录，文�g名都�?empty_msg.proto.
�q�行时就会报错：

[libprotobuf ERROR E:\deps\protobuf-3.2.0\protobuf-3.2.0\src\google\protobuf\
descriptor_database.cc:57] File already exists in database : empty_msg.proto
[libprotobuf FATAL E:\deps\protobuf-3.2.0\protobuf-3.2.0\src\google\protobuf\
descriptor.cc:1275] CHECK failed: generated_database_->Add(encoded_file_descriptor, size):

原因��囄��同一个文件名"empty_msg.proto"往descriptor_database��d��descriptor�?br />
如果按包名分子目录，文�g名就可以分开�?"rpc/empty_msg.proto" �?"svr/empty_msg.proto".

金庆 2017-04-17 14:40 发表评论

behaviac动态库�q�行出错

金庆 — Thu, 16 Mar 2017 03:40:00 GMT

behaviac动态库�q�行出错

(金庆的专�?2017.3.16)

游戏是静态链接的�q�行库，��d��behaviac动态库后，�q�行出错�Q?br />
>   ucrtbased.dll!free_dbg_nolock(void * const block, const int block_use) �?996   C++
    ucrtbased.dll!_free_dbg(void * block, int block_use) �?1030   C++
    libbehaviac_msvc_debug.dll!operator delete(void * block) �?17   C++
    libbehaviac_msvc_debug.dll!std::_Deallocate(void * _Ptr, unsigned int _Count, unsigned int _Sz) �?132   C++
    libbehaviac_msvc_debug.dll!std::allocator::deallocate(std::_Container_proxy * _Ptr, unsigned int _Count) �?720   C++
    libbehaviac_msvc_debug.dll!std::_Wrap_alloc >::deallocate(std::_Container_proxy * _Ptr, unsigned int _Count) �?988   C++
    libbehaviac_msvc_debug.dll!std::_String_alloc > >::_Free_proxy() �?661   C++
    libbehaviac_msvc_debug.dll!std::_String_alloc > >::~_String_alloc > >() �?629   C++
    libbehaviac_msvc_debug.dll!std::basic_string,std::allocator >::~basic_string,std::allocator >() �?1018   C++
    libbehaviac_msvc_debug.dll!behaviac::CTagObject::Save(behaviac::IIONode * node, const char * szClassName) �?100   C++
    libbehaviac_msvc_debug.dll!behaviac::Agent::InitVariableRegistry() �?476   C++
    libbehaviac_msvc_debug.dll!behaviac::Agent::Init_(int contextId, behaviac::Agent * pAgent, short priority, const char * agentInstanceName) �?227   C++
    giant_test_client.exe!behaviac::Agent::InitAgent(behaviac::Agent * pAgent, const char * agentInstanceName, const char * agentInstanceNameAny, bool bToBind, int contextId, short priority) �?56   C++
    giant_test_client.exe!behaviac::Agent::Create(const char * agentInstanceName, int contextId, short priority) �?88   C++
    giant_test_client.exe!Client::InitPlayer() �?36   C++
    giant_test_client.exe!Client::Client() �?23   C++
    giant_test_client.exe!main() �?50   C++
    giant_test_client.exe!invoke_main() �?64   C++
    giant_test_client.exe!__scrt_common_main_seh() �?253   C++
    giant_test_client.exe!__scrt_common_main() �?296   C++
    giant_test_client.exe!mainCRTStartup() �?17   C++
    kernel32.dll!@BaseThreadInitThunk@12()   未知
    ntdll.dll!___RtlUserThreadStart@8()   未知
    ntdll.dll!__RtlUserThreadStart@8()   未知

看来是string跨越动态库析构�?br />如果string在动态库中构造又析构�Q�动态库的运行时库可以与�ȝ��序不同�?br />但如果用�ȝ��序的�q�行时申请内存，在动态库中用另一个运行时库释放，��׃��产生上面的错误�?br />
可以建立一个最��化的程序来复现上面的错误�?br />新徏一个行为树�Q��ؓAgent创徏一个属性�?br />然后�ȝ��序运行库设�ؓ�Q�多�U�程调试 (/MTd)�Q�而不是：多线�E�调�?DLL (/MDd)�?br />调试�q�行创徏Agent时就会出上面错误�?

疑问是这是动态库内部调用�Q�怎么会��生跨库析构呢�Q?br />
断点�q�入string的构造过�E�可以发玎ͼ�string不是动态库构造的�Q�而是�ȝ��序构造的�Q?br />
>   test_client.exe!behaviac::StringUtils::internal::ToString(const std::basic_string,std::allocator > & val) �?138   C++
    test_client.exe!behaviac::StringUtils::Detail::ToStringPtrHanler,std::allocator >,0>::ToString(const std::basic_string,std::allocator > & v) �?192   C++
    test_client.exe!behaviac::StringUtils::Detail::ToStringEnumHanler,std::allocator >,0>::ToString(const std::basic_string,std::allocator > & v) �?228   C++
    test_client.exe!behaviac::StringUtils::Detail::ToStringStructHanler,std::allocator >,0>::ToString(const std::basic_string,std::allocator > & val) �?242   C++
    test_client.exe!behaviac::StringUtils::ToString,std::allocator > >(const std::basic_string,std::allocator > & val) �?256   C++
    test_client.exe!behaviac::SetFromString_t,std::allocator >,0>::Get(std::basic_string,std::allocator > & value) �?790   C++
    test_client.exe!behaviac::CProperty,std::allocator > >::GetValueToString(const behaviac::Agent * self) �?848   C++
    libbehaviac_msvc_debug.dll!behaviac::CTagObject::Save(behaviac::IIONode * node, const char * szClassName) �?97   C++
    libbehaviac_msvc_debug.dll!behaviac::Agent::InitVariableRegistry() �?476   C++
    libbehaviac_msvc_debug.dll!behaviac::Agent::Init_(int contextId, behaviac::Agent * pAgent, short priority, const char * agentInstanceName) �?227   C++
    test_client.exe!behaviac::Agent::InitAgent(behaviac::Agent * pAgent, const char * agentInstanceName, const char * agentInstanceNameAny, bool bToBind, int contextId, short priority) �?56   C++
    test_client.exe!behaviac::Agent::Create(const char * agentInstanceName, int contextId, short priority) �?88   C++
    test_client.exe!Client::InitPlayer() �?36   C++
    test_client.exe!Client::Client() �?23   C++
    test_client.exe!main() �?50   C++
    test_client.exe!invoke_main() �?64   C++
    test_client.exe!__scrt_common_main_seh() �?253   C++
    test_client.exe!__scrt_common_main() �?296   C++
    test_client.exe!mainCRTStartup() �?17   C++
    kernel32.dll!75c3336a()   未知
    [下面的框架可能不正确�?或缺失，没有�?kernel32.dll 加蝲�W�号]
    ntdll.dll!76fa9902()   未知
    ntdll.dll!76fa98d5()   未知

原因�?IProperty::GetValueToString()是虚函数�Q�实际实现是�׃��E�序实现模板�c�R�?br />   behaviac::CProperty<>::GetValueToString()

解决�Ҏ��Z��用behaviac静态库�q�且静态链接运行时库�?br />

金庆 2017-03-16 11:40 发表评论

Lua53 premake

金庆 — Sat, 18 Feb 2017 14:18:00 GMT

Lua53 premake

(金庆的专�?2017.2)

参考：用premake5创徏lua532工程
      http://blog.csdn.net/jq0123/article/details/51242780

-- premake5.lua
--[[
Usage examples:
   for windows: premake5.exe --os=windows vs2015
   fot linux:   premake5.exe --os=linux gmake
]]

workspace "lua53"
   configurations { "Debug", "Release" }
   targetdir "bin/%{cfg.buildcfg}"

   language "C++"
   -- Force VS to compile as C++.
   -- https://github.com/premake/premake-core/issues/142
   filter "action:vs*"
      buildoptions "/TP"

   filter "system:windows"
      defines { "LUA_BUILD_AS_DLL" }

   filter "configurations:Debug"
      defines { "DEBUG" }
      flags { "Symbols" }

   filter "configurations:Release"
      defines { "NDEBUG" }
      optimize "On"

project "lua53"
   kind "ConsoleApp"
   files { "src/lua.c" }
   links { "lua53_shared_lib" }

project "luac53"
   kind "ConsoleApp"
   files { "src/luac.c" }
   links { "lua53_static_lib" } -- Link error on Windows if link lua53 shared lib.

project "lua53_shared_lib"
   kind "SharedLib"
   targetname "lua53"
   files { "src/*.h", "src/*.c" }
   removefiles { "src/lua.c", "src/luac.c" }

project "lua53_static_lib"
   kind "StaticLib"
   targetname "lua53"
   filter "system:windows"
      targetprefix "lib" -- liblua53.lib
   filter {}
   files { "src/*.h", "src/*.c" }
   removefiles { "src/lua.c", "src/luac.c" }

更改之处�Q?br />* VS强制按C++�~�译
* 创徏动态库和静态库
* lua53.exe 链接动态库�Q�luac53.exe 链接静态库

因�ؓ luac53.exe 链接动态库�~?个函数未导出�?br />* ��d��?LUA_BUILD_AS_DLL�Q�不�?lua53.dll 不会生成 lua53.lib

金庆 2017-02-18 22:18 发表评论

用grpc_cb代替grpc++

金庆 — Sun, 22 Jan 2017 10:06:00 GMT

用grpc_cb代替grpc++

(金庆的专�?2017.1)

jinq0123/grpc_cb
�?Google gRpc 的C++库�?
它依赖于 grpc, 采用回调接口�Q�简化了使用�Q�用来代�?grpc++ 库�?/p>

使用��介如下�?/p>

定义服务

�?proto 文�g定义服务�Q?/p>

// See examples/protos/route_guide.proto.

syntax = "proto3";

package routeguide;

// Interface exported by the server.
service RouteGuide {
// A simple RPC.
rpc GetFeature(Point) returns (Feature) {}
}

message Point {
int32 latitude = 1;
int32 longitude = 2;
}

message Feature {
string name = 1;
Point location = 2;
}

生成服务器和客户端代�?p>

详见�Q�examples/protos/generate.bat

protoc.exe -I . --cpp_out=../cpp_cb/route_guide route_guide.proto
protoc.exe -I . --grpc_out=../cpp_cb/route_guide --plugin=protoc-gen-grpc=grpc_cpp_cb_plugin.exe route_guide.proto

在examples/cpp_cb/route_guide/ 生成以下文�g:

route_guide.pb.h, 消息�c�d��?/li>
route_guide.pb.cc, 消息�c�d��?/li>
route_guide.grpc_cb.pb.h, 服务�c�d��?/li>
route_guide.grpc_cb.pb.cc, 服务�c�d��?/li>

生成的命名空�?code>RouteGuide��包�?/p>

客户端��用的Stub�c?
需服务器实现的Service�c?

客户端调用RPC

同步调用

ChannelSptr channel(new Channel("localhost:50051"));
Stub stub(channel);

Point point = MakePoint(0, 0);
Feature feature;
Status status = stub.BlockingGetFeature(point, &feature);

异步调用

stub.AsyncGetFeature(point,
	[](const Feature& feature) {
		cout << feature.name() << endl;
	});

服务器实�?/h2>

先实现服务类

    class RouteGuideImpl final : public routeguide::RouteGuide::Service {
   public:
       void GetFeature(const Point& point,
               const GetFeature_Replier& replier) override {
           Feature feature;
           feature.set_name("...");
           replier.Reply(feature);
       }
   }

GetFeature()不必立即应答�Q�可复制保存replier后直接返回， 待应�{�内容准备完成后�Q�再调用Reply().

开启服�?/pre>

    Server svr;
   svr.AddListeningPort("0.0.0.0:50051");
   RouteGuideImpl service(db_path);
   svr.RegisterService(service);
   svr.BlockingRun();

金庆 2017-01-22 18:06 发表评论

hiredis异步接口��装�q�导出到Lua

金庆 — Thu, 05 Jan 2017 10:42:00 GMT

hiredis异步接口��装�q�导出到Lua

(金庆的专�?2017.1)

hiredis 不支�?Windows, Windows 下��?wasppdotorg / hiredis-for-windows �?br />Linux 下仍�?redis/hiredis�?br />hiredis-for-windows 是以 hiredis 0.13.3 为基��U�L��的�?br />
hiredis-for-windows 需要稍加修正：
   * 去除 inline �?br />   * TCP_NODELAY 改在�q�接之前讄��?br />详见其Issue.

Cluster 支持采用 shinberg/cpp-hiredis-cluster。这是个CPP库，支持异步�Q?br />要求 hiredis >= 0.12.0�?br />jinq0123/cpp-hiredis-cluster �?develop 分支上更改了接口�Q�让它更好用�?br />
因�ؓ�|�络库是boost asio, 所以需要asio适配器，采用 jinq0123/hiredis-boostasio-adapter�?br />
cpp-hiredis-cluster 提供的是�l�一的Command接口�Q�接收字�W�串命��o�Q�返�?redisReply.
对于常用命��o�Q�需要更��单的接口�?br />在Lua手游服务器代码中新徏CRedis�c�，��装 cpp-hiredis-cluster�Q?br />为常用redis命��o��装更好用的接口�?br />CRedis�c�d��装了asio, 接口是根据应用需要定义的�Q�所以是专用接口�Q?br />不在 cpp-hiredis-cluster 中实现�?br />
bool CRedis::Init(io_service& rIos, const std::string& sHost, uint16_t uPort)
创徏 RedisCluster 对象�?br />io_service 用于创徏一�?redis 事�g适配器，
RedisCluster创徏需要一个适配器�?br />sHost, uPort 用于初始化连接redis cluster, 获取集群信息�?br />

using Cmd = RedisCluster::AsyncHiredisCommand;

bool CRedis::Init(io_service& rIos, const std::string& sHost, uint16_t uPort)
{
   m_pAdapter.reset(new Adapter(rIos));
   try
   {
       m_pCluster.reset(Cmd::createCluster("127.0.0.1", 7000, *m_pAdapter));
   }
   catch (const RedisCluster::ClusterException &e)
   {
       LOG_ERROR("Cluster exception: " << e.what());
       return false;
   }

   return true;
}

static Cmd::Action handleException(const RedisCluster::ClusterException &exception,
   RedisCluster::HiredisProcess::processState state)
{
   // Check the exception type.
   // Retry in case of non-critical exceptions.
   if (!dynamic_cast(&exception))
   {
       LOG_WARN("Exception in processing async redis callback: "
           << exception.what() << " Retry...");
       // retry to send a command to redis node
       return Cmd::RETRY;
   }
   LOG_ERROR("Critical exception in processing async redis callback: "
       << exception.what());
   return Cmd::FINISH;
}

static void handleSetReply(const redisReply &reply, const CRedis::SetCb& setCb)
{
   if (!setCb) return;
   if (reply.type == REDIS_REPLY_STATUS)
   {
       const std::string OK("OK");
       if (OK == reply.str)
       {
           setCb(true);
           return;
       }
   }

   LOG_WARN("Set reply: " << reply.str);
   setCb(false);
}

void CRedis::Set(const string& sKey, const string& sValue, const SetCb& setCb)
{
   assert(m_pCluster);
   Cmd::commandf2(*m_pCluster, sKey,
       [setCb](const redisReply& reply) {
           handleSetReply(reply, setCb);
       },
       handleException,
       "set %s %s", sKey.c_str(), sValue.c_str());
}

static void handleGetReply(const redisReply& reply,
   const CRedis::ReplyStringCb& hdlStrReply)
{
   if (!hdlStrReply) return;
   using Rt = CRedis::ReplyType;
   if (reply.type == REDIS_REPLY_NIL)
   {
       hdlStrReply(Rt::NIL, "");
       return;
   }
   std::string sReply(reply.str, reply.len);
   if (reply.type == REDIS_REPLY_STRING)
       hdlStrReply(Rt::OK, sReply);
   else
       hdlStrReply(Rt::ERR, sReply);
}

void CRedis::Get(const std::string& sKey, const ReplyStringCb& hdlStrRepl)
{
   assert(m_pCluster);
   Cmd::commandf2(*m_pCluster, sKey,
       [hdlStrRepl](const redisReply& reply) {
           handleGetReply(reply, hdlStrRepl);
       },
       handleException,
       "get %s", sKey.c_str());
}

handleException 是Cmd::command() 接口中需要的异常处理�Q�这里是��量重试�?br />
Cmd::command() 中的�W?个参数是 redis 应答的回调，��d�� redisReply, 然后触发命��o的回调�?br />
CRedis::Get() 执行 redis GET 命��o�Q�固�?个参敎ͼ��q�回是字�W�串�Q�nil, 或错误�?br />
   enum class ReplyType
   {
       OK = 0, // redis replys string/integer/array
       NIL = 1, // redis replys nil
       ERR = 2, // redis replys error status
   };

   // ��单的常用命��o会自动解析reply, 使用更易用的回调�?/span>
   using ReplyStringCb = function;

CRedis::Get() 的回调就�?ReplyStringCb�?br />
void CRedis::Set() 的回调只需知道成功或失败，SetCb 定义为：
   using SetCb = function;

��p�|时会有错误信息，已统一打印日志�Q�不再暴露出来�?br />
SET 命��o完整的参数列表还有其他参敎ͼ�
set key value [EX seconds] [PX milliseconds] [NX|XX]

因�ؓ常用的就 "set key value", 其他扩展的命令需要��用通用�?command() 接口�Q?br />�q��要读�?redisReply.

使用 LuaIntf 导出 CRedis �?Lua.
因�ؓ只有一�?CRedis, 所以导��Zؓ模块 c_redis:

#include

namespace {

void Set(const string& sKey, const string& sValue, const LuaRef& luaSetCb)
{
   // Default is empty callback.
   auto setCb = ToFunction(luaSetCb);
   GetRedis().Set(sKey, sValue, setCb);
}

void Get(const string& sKey, const LuaRef& luaReplyStringCb)
{
   auto replyStringCb = ToFunction(
       luaReplyStringCb);
   GetRedis().Get(sKey, replyStringCb);
}

} // namespace

void Bind(lua_State* L)
{
   LuaBinding(L).beginModule("c_redis")
       .addFunction("set", &Set)
       .addFunction("get", &Get)
   .endModule();
}

需要将 lua 的回调函数�{�?cpp 的回调：

template
Function ToFunction(const LuaIntf::LuaRef& luaFunction)
{
   // Default is empty.
   if (!luaFunction)
       return Function(); // skip nil
   if (luaFunction.isFunction())
       return luaFunction.toValue(); // Todo: catch
   LOG_WARN_TO("ToFunction", "Lua function expected, but got "
       << luaFunction.typeName());
   return Function();
}

Lua �q�样调用�Q?br />c_redis.set("FOO", "1234")
c_redis.set("FOO", "1234", function(ok) print(ok) end)
c_redis.get("FOO", function(reply_type, reply)
   assert("string" == type(reply))
   if 0 == reply_type or 1 == reply_type then
       print("FOO="..reply)
   else
       print("Error: "..reply)
   end
end)

金庆 2017-01-05 18:42 发表评论

hiredis的各�U�windows版本

金庆 — Wed, 28 Dec 2016 03:02:00 GMT

hiredis的各�U�windows版本

(金庆的专�?2016.12)

hiredis 是内存数据库 redis 的客��L��C�? 不支持Windows�?br />
hiredis的Windows�U�L��版本有许多：

desb42/hiredis
   forked from redis/hiredis
   hiredis 0.10.1
   Star 3

koenvandesande/hiredis
   forked from redis/hiredis
   hiredis 0.11.0
   在日志中指出��Z�� desb42
       Windows compatability, partially based on desb42's patch, but with cleanup and additional fixes.
   Star 8

wasppdotorg/hiredis-for-windows
   hiredis 0.13.3
   README.md 头部指出��Z�� koenvandesande/hiredis
       https://github.com/redis/hiredis (0.13.3)
       https://github.com/koenvandesande/hiredis
   Star 3

lgsonic/hiredis-win
   hiredis 0.10.1
   Star 15

texnician/hiredis-win32
   hiredis 0.10.1
   Star 17

Microsoft/hiredis
   forked from redis/hiredis
   hiredis 0.11.0
   Star 11

ayrb13/hiredis-win
   hiredis 0.11.0
   Star 1

对于星星数都较少的情况，创徏较早�?hiredis-win �?hiredis-win32 星星��C��占据优势�Q?br />但是星星意义不大�?br />
支持hiredis的版本是关键�?br />hiredis-for-windows 支持版本最斎ͼ��q�且渊源清楚�Q�可以信赖�?br />
Microsoft/hiredis ��着MS的牌子比较容易让人接受，因�ؓ�?forked from redis/hiredis�Q?br />所以升�U?hiredis 只需处理下冲�H�就行了�?br />但是�q�个�U�L��更改太大�Q��用了IOCP, 多了一个win32_interop, 不如其他�U�L��z��?br />�q�应该是 MSOpenTech/redis 的子��目�Q�见�Q?br />http://blog.sina.com.cn/s/blog_47379bd80102vbtb.html
Win32_Interop 重定义了一些Windows API以模拟Linux下的POSIX函数�?br />在项目中�q�接hiredis.lib和Win32_Interop.lib�Ӟ��如果同时�q�接�pȝ��库文�Ӟ��
则会出现一�p�d��冲突�?br />

金庆 2016-12-28 11:02 发表评论

�l�构体初始化列表错误

金庆 — Mon, 12 Dec 2016 09:16:00 GMT

�l�构体初始化列表错误

(金庆的专�?2016.12)

struct A
{
   int a = 0;
};

int main()
{
   A a{0};
   return 0;
}

报以下错误：

error C2440: “初始�?#8221;: 无法�?#8220;initializer list”转换�?#8220;A”
note: 无构造函数可以接受源�c�d��Q�或构造函数重载决�{�不明确

去除 A.a 的类内初始化��好了�?br />
struct A
{
   int a;
};

应该是添加类内初始化后，��׃��再有默认构造函��C��?br />

金庆 2016-12-12 17:16 发表评论

C++用LuaIntf调用Lua代码�C�Z��

金庆 — Fri, 09 Dec 2016 14:17:00 GMT

C++用LuaIntf调用Lua代码�C�Z��

(金庆的专�?2016.12)

    void LuaTest::OnResponse(uint32_t uLuaRpcId,
       const std::string& sRespContent) const
   {
       using LuaIntf::LuaRef;
       LuaRef require(m_pLuaState, "require");
       try {
           LuaRef handler = require.call("client_rpc_response_handler");
           handler.dispatchStatic("handle", uLuaRpcId, sRespContent);
       }
       catch (const LuaIntf::LuaException& e) {
           std::cerr << "Failed to call lua client_rpc_response_handler.handle(), "
               << e.what() << std::endl;
       }
   }

�q�是��试客户端代码，可以写Lua代码��试服务器．

Lua代码中发��Z��个Rpc��h��? 会在Lua中保存一个回�? 待收到应�{�时触发回调. 通过uLuaRpcId来烦引该回调.

sRespContent 是收到的应答�? ��在lua中解�?

OnResponse() ��是调用�?Lua 代码:

    require("client_rpc_response_handler").handle(uLuaRpcId, sRespContent)

金庆 2016-12-09 22:17 发表评论

log4cxx用环境变量设�|�输出文件名

金庆 — Mon, 05 Dec 2016 07:31:00 GMT

log4cxx用环境变量设�|�输出文件名

(金庆的专�?2016.12)

利用环境变量�Q�可以用同一个log4j.xml来配�|�多个相��D��E�，输出日志��C��同文件�?br />
例如多个BaseApp�q�程使用同一个BaseApp.xml配置, SERVER_ID为环境变量：


      ...

代码启动时先��d��server_id参数�Q�然后设�|?SERVER_ID 环境变量�Q�然后再配置log4cxx.

int main(int argc, char* argv[])
{
   log4cxx::NDC ndcMain("");
   if (argc < 3)
   {
       LOG_ERROR(Fmt("Usage: %s cfg_file server_id") % argv[0]);
       return -1;
   }
   uint16_t uServerId = (uint16_t)atoi(argv[2]);
   if (!Util::SetServerIdEnv(uServerId)) // for log4cxx
       return -1;

   // Must after SetServerIdEnv().
   log4cxx::xml::DOMConfigurator::configureAndWatch("log4j/BaseApp.xml", 5000);
   LOG_INFO("--------------------------- ");
   LOG_INFO(Fmt("Start base app (ID=%1%).") % uServerId);
   LOG_INFO("--------------------------- ");
   ...
}

SetServerIdEnv() 如下�Q?br />
bool SetServerIdEnv(uint16_t uServerId)
{
   const char LOG_NAME[] = "SetServerIdEnv";
   static char buf[128] = {0}; // putenv need a buffer
   int nLen = snprintf(buf, sizeof(buf), "SERVER_ID=%u", uServerId);
   if (nLen < 0)
   {
       LOG_ERROR(Fmt("snprintf() failed. (%1%)%2%") % errno % strerror(errno));
       return false;
   }

   int nErr = putenv(buf);
   if (0 == nErr) return true;
   LOG_ERROR(Fmt("putenv() failed. (%1%)%2%") % errno % strerror(errno));
   return false;
}

�q�行目录下有个log4cxx�~�省配置 log4j.xml, 会自动加载，
所以在 log4cxx 昑ּ�配置之前也可以调用日志输出�?br />
�q�行多个BaseApp.exe:

start Debug\Giant_BaseApp.exe cfg.ini 4
start Debug\Giant_BaseApp.exe cfg.ini 3

金庆 2016-12-05 15:31 发表评论

金庆 — Fri, 18 Nov 2016 08:19:00 GMT

gdb不知��Z��昄��2�ơ析�?br />
(金庆的专�?2016.11)

gdb 昄��2��?A::~A():

    (gdb) bt
    #0 A::~A (this=0x602010, __in_chrg=) at main.cpp:10
    #1 0x0000000000400a96 in A::~A (this=0x602010, __in_chrg=)
        at main.cpp:12
    #2 0x00000000004009c0 in main () at main.cpp:18

代码如下�Q?br />
    class A
    {
    public:
        A() {}
        virtual ~A()
        {
            cout << "~A()" << endl;
        }
    };

    int main()
    {
        A* p = new A;
        delete p;
        return 0;
    }

打断�Ҏ��C�：2 locations:

    (gdb) b A::~A
    Breakpoint 1 at 0x400a40: A::~A. (2 locations)

完整�?gdb 输出:

    g++ -g main.cpp
    [jinq@localhost test]$ gdb a.out
    GNU gdb (GDB) Red Hat Enterprise Linux 7.6.1-80.el7
    Copyright (C) 2013 Free Software Foundation, Inc.
    License GPLv3+: GNU GPL version 3 or later
    This is free software: you are free to change and redistribute it.
    There is NO WARRANTY, to the extent permitted by law. Type "show copying"
    and "show warranty" for details.
    This GDB was configured as "x86_64-redhat-linux-gnu".
    For bug reporting instructions, please see:
    ...
    Reading symbols from /home/jinq/test/a.out...done.
    (gdb) b A::~A
    Breakpoint 1 at 0x400a40: A::~A. (2 locations)
    (gdb) run
    Starting program: /home/jinq/test/a.out

    Breakpoint 1, A::~A (this=0x602010, __in_chrg=) at main.cpp:12
    12          }
    Missing separate debuginfos, use: debuginfo-install glibc-2.17-106.el7_2.8.x86_64 libgcc-4.8.5-4.el7.x86_64 libstdc++-4.8.5-4.el7.x86_64
    (gdb) bt
    #0 A::~A (this=0x602010, __in_chrg=) at main.cpp:12
    #1 0x00000000004009c0 in main () at main.cpp:18
    (gdb) s

    Breakpoint 1, A::~A (this=0x602010, __in_chrg=) at main.cpp:10
    10          {
    (gdb) bt
    #0 A::~A (this=0x602010, __in_chrg=) at main.cpp:10
    #1 0x0000000000400a96 in A::~A (this=0x602010, __in_chrg=)
        at main.cpp:12
    #2 0x00000000004009c0 in main () at main.cpp:18
    (gdb) ^CQuit
    (gdb)

虚析构函数�ƈ且是delete才这栗��?br />
Also see: http://lists.qt-project.org/pipermail/interest/2015-November/019691.html

已提交Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=20837

金庆 2016-11-18 16:19 发表评论

金庆 — Wed, 28 Sep 2016 09:44:00 GMT

gloox配置聊天�?br />
(金庆的专�?

gloox是XMPP协议的C++客户端库�?br />以下代码��试创徏多�h聊天�?MUC), �q�进行配�|��?br />参照gloox中的muc�C�Z��代码�?br />gloox代码�C�Z��中没有聊天室的配�|��?br />配置聊天室需要获取配�|�表�?DataForm), 填好表单然后调用 setRoomConfig().
配置表单��d��服务器发来的默认配置�Q�仅更改了其中一��V�?br />��试服务器��用了ejabberd.

const char SERVER[] = "xmpp.jinqing.net";
const char TESTER[] = "tester";
const char PASSWORD[] = "password";

using namespace gloox;

static std::string GetTesterJid()
{
   return std::string(TESTER) + "@" + SERVER;
}

static DataForm* CreateMUCConfigForm(const DataForm& form)
{
   DataForm* pNewFm = new DataForm(TypeSubmit);
   const DataForm::FieldList& fl = form.fields();
   for (const DataFormField* pFld : fl)
   {
       DataFormField* pNewFld = pNewFm->addField(
           pFld->type(), pFld->name(), pFld->value());
       if (pFld->name() == "muc#roomconfig_roomdesc")
           pNewFld->setValue("RoomDesc_JinqTest");
   }
   return pNewFm;
}

class MUCRoomConfigHandlerTest : public MUCRoomConfigHandler
{
public:
   void handleMUCConfigList(MUCRoom* room, const MUCListItemList& items,
       MUCOperation operation) override {}

   void handleMUCConfigForm(MUCRoom* room, const DataForm& form) override
   {
       DataForm* pNewForm = CreateMUCConfigForm(form); // deleted in setRoomConfig()
       room->setRoomConfig(pNewForm);
   }

   void handleMUCConfigResult(MUCRoom* room, bool success,
       MUCOperation operation) override {}
   void handleMUCRequest(MUCRoom* room, const DataForm& form) override {}
}; // class MUCRoomConfigHandlerTest

class CreateRoomTest : public ConnectionListener, MUCRoomHandler
{
public:
   CreateRoomTest() : m_client(JID(GetTesterJid()), PASSWORD)
   {
       m_client.registerConnectionListener(this);
   }

public:
   void TestCreate()
   {
       m_client.setPresence(Presence::Available, -1);
       JID nick(std::string("gloox@conference.") + SERVER + "/gloox");
       m_pRoom.reset(new MUCRoom(&m_client, nick, this, &m_cfgHdlr));
       m_client.connect();
   }

   void onConnect() override
   {
       m_pRoom->join();
       m_pRoom->requestRoomConfig();
   }

   void onDisconnect(ConnectionError /*e*/) override {}

   bool onTLSConnect(const CertInfo& info) override
   {
       return true;
   }

   void handleMUCParticipantPresence(MUCRoom * /*room*/,
       const MUCRoomParticipant participant,
       const Presence& presence) override {}

   void handleMUCMessage(MUCRoom* /*room*/,
       const Message& msg, bool priv) override {}

   void handleMUCSubject(MUCRoom * /*room*/, const std::string& nick,
       const std::string& subject) override {}

   void handleMUCError(MUCRoom * /*room*/, StanzaError error) override {}

   void handleMUCInfo(MUCRoom * /*room*/, int features,
       const std::string& name, const DataForm* infoForm) override {}

   void handleMUCItems(MUCRoom * /*room*/,
       const Disco::ItemList& items) override {}

   void handleMUCInviteDecline(MUCRoom * /*room*/, const JID& invitee,
       const std::string& reason) override {}

   bool handleMUCRoomCreation(MUCRoom *room) override
   {
       return true;
   }

private:
   gloox::Client m_client;
   MUCRoomConfigHandlerTest m_cfgHdlr;
   std::unique_ptr m_pRoom;
}; // class CreateRoomTest

int main()
{
   CreateRoomTest().TestCreate();
   return 0;
}

金庆 2016-09-28 17:44 发表评论

天堂久久天堂AV色综合,久久久www免费人成精品,久久性生大片免费观看性

How are dtLinks created in NavMesh

How are dtLinks created in NavMesh

dtLink

How link is used

fields of dtLink

How are links created

1. Connect internal links

2. Base off-mesh connections

3. Connect external off-mesn links within this tile

4. Connect with layers in current tile

5. Connect with neighbour tiles

Reference

C++ parameter passing rules

C++ parameter passing rules

Naming Conventions for Accessors

Naming Conventions for Accessors

Visual Studio 2019 Compiler Hangs

Fbx File Format Identifier

查找内存错误

std::thread 中的异常会丢��p��用栈

用赋��g���?protobuf CopyFrom()

vs2017 linux �~�译输出�Ҏ�� vs 格式

�?LiteIDE ��d��选中标记

asio 协程�?yield

Lua和C++之间调用效率���试

conan-transit服上的库列表

VC6工程因行���格式无法�{换到VS2015

std::hash实现太简单分布不匀

用preload加蝲Lua导出模块

支持 proto3 �?lua �l�定�?LuaPbIntf

behaviac动态库�q�行出错

Lua53 premake

用grpc_cb代替grpc++

用grpc_cb代替grpc++

定义服务

客户端调用RPC

同步调用

异步调用

hiredis异步接口���装�q�导出到Lua

hiredis的各�U�windows版本

�l�构体初始化列表错误

C++用LuaIntf调用Lua代码�C�Z��

C++用LuaIntf调用Lua代码�C�Z��

log4cxx用环境变量设�|�输出文件名

`dtLink`

fields of `dtLink`

用赋��g��?protobuf CopyFrom()

Lua和C++之间调用效率��试

VC6工程因行��格式无法�{换到VS2015

hiredis异步接口��装�q�导出到Lua