blob: d267333657b77680c85414cbf195005bd4b31b0b [file] [log] [blame]
<?xml version="1.0" standalone="no"?>
<!--
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
-->
<!DOCTYPE s1 SYSTEM "sbk:/style/dtd/document.dtd">
<s1 title="Build Instructions">
<s2 title="Build Instructions">
<p>Build instructions are provided for the following platforms and
compilers:</p>
<ul>
<li><link anchor="CMake">All platforms</link></li>
<li><link anchor="UNIX">UNIX/Linux/Mac OS X/Cygwin/MinGW</link></li>
</ul>
<anchor name="CMake"/>
<s3 title="Building on all platforms with CMake">
<p>For building on any platform with any supported build
system &XercesCName; uses the CMake build generator and
requires that you have <jump
href="https://cmake.org/">CMake</jump> installed.
Additionally, a build tool such as <jump
href="http://www.gnu.org/software/make/make.html">GNU
make</jump> or <jump
href="https://ninja-build.org/">Ninja</jump> is required for
building. CMake supports a wide range of generators for
several different compilers, build tools and popular IDEs,
including Eclipse, Kate, Visual Studio, Sublime Text and more.
Any of these may be used to build &XercesCName;. Run
<code>cmake --help</code> to display the full list of
supported generators for your platform.</p>
<p>As with all CMake projects, the build process is divided
into several parts: configuration and building, followed by
(optional) testing and installation. The configuration part is
performed by running the <code>cmake</code> command. The
build part is performed by invoking the chosen build tool
such as <code>make</code> or <code>ninja</code>, or by opening
the generated project files in your IDE, and building from
within the IDE.</p>
<p>Besides the standard <code>cmake</code> <jump
href="https://cmake.org/cmake/help/latest/manual/cmake-variables.7.html">variables</jump>,
&XercesCName; provides a number of project-specific options
that are worth mentioning. You can specify one option for each
category outlined below. If you do not specify anything for a
particular category then <code>cmake</code> will select the
most appropriate default, based upon the available options for
your system. At the end of its execution <code>cmake</code>
prints the selected values for each category.</p>
<p>Net Accessor (used to access network resources):</p>
<table>
<tr>
<th>Option</th>
<th>Description</th>
</tr>
<tr>
<td><code>-Dnetwork-accessor=curl</code></td>
<td>use the libcurl library (only on UNIX)</td>
</tr>
<tr>
<td><code>-Dnetwork-accessor=socket</code></td>
<td>use plain sockets (only on UNIX)</td>
</tr>
<tr>
<td><code>-Dnetwork-accessor=cfurl</code></td>
<td>use the CFURL API (only on Mac OS X)</td>
</tr>
<tr>
<td><code>-Dnetwork-accessor=winsock</code></td>
<td>use WinSock (only on Windows, Cygwin, MinGW)</td>
</tr>
<tr>
<td><code>-Dnetwork:BOOL=OFF</code></td>
<td>disable network support</td>
</tr>
</table>
<p>Transcoder (used to convert between internal UTF-16 and other encodings):</p>
<table>
<tr>
<th>Option</th>
<th>Description</th>
</tr>
<tr>
<td><code>-Dtranscoder=gnuiconv</code></td>
<td>use the GNU iconv library</td>
</tr>
<tr>
<td><code>-Dtranscoder=iconv</code></td>
<td>use the iconv library</td>
</tr>
<tr>
<td><code>-Dtranscoder=icu</code></td>
<td>use the ICU library</td>
</tr>
<tr>
<td><code>-Dtranscoder=macosunicodeconverter</code></td>
<td>use Mac OS X APIs (only on Mac OS X)</td>
</tr>
<tr>
<td><code>-Dtranscoder=windows</code></td>
<td>use Windows APIs (only on Windows and MinGW)</td>
</tr>
</table>
<p>Message Loader (used to access diagnostics messages):</p>
<table>
<tr>
<th>Option</th>
<th>Description</th>
</tr>
<tr>
<td><code>-Dmessage-loader=inmemory</code></td>
<td>store the messages in memory</td>
</tr>
<tr>
<td><code>-Dmessage-loader=icu</code></td>
<td>store the messages using the ICU resource bundles</td>
</tr>
<tr>
<td><code>-Dmessage-loader=iconv</code></td>
<td>store the messages in the iconv message catalog</td>
</tr>
</table>
<p>XMLCh type (UTF-16 character type):</p>
<table>
<tr>
<th>Option</th>
<th>Description</th>
</tr>
<tr>
<td><code>-Dxmlch-type=char16_t</code></td>
<td>use <code>char16_t</code> (requires a C++11 compiler)</td>
</tr>
<tr>
<td><code>-Dxmlch-type=uint16_t</code></td>
<td>use <code>uint16_t</code> from
<code>&lt;cstdint&gt;</code> or
<code>&lt;stdint.h&gt;</code>, or another unsigned
16-bit type such as <code>unsigned short</code> if
the standard types are unavailable</td>
</tr>
<tr>
<td><code>-Dxmlch-type=wchar_t</code></td>
<td>use <code>wchar_t</code> (Windows only)</td>
</tr>
</table>
<p>MFC debug support is enabled by default (Windows only) and
can be disabled with the
<code>-Dmfc-debug:BOOL=OFF</code> option.</p>
<p>Thread support is enabled by default and can be disabled
with the <code>-Dthreads:BOOL=OFF</code> option. If disabled,
it will not be possible to select a mutex manager other than
<code>nothreads</code>. If enabled, one of the following
mutex managers may be selected:</p>
<table>
<tr>
<th>Option</th>
<th>Description</th>
</tr>
<tr>
<td><code>-Dmutex-manager=standard</code></td>
<td>Use Standard C++ mutex (requires a C++11 compiler)</td>
</tr>
<tr>
<td><code>-Dmutex-manager=posix</code></td>
<td>Use POSIX threads (pthreads) mutex (only on UNIX and Cygwin)</td>
</tr>
<tr>
<td><code>-Dmutex-manager=windows</code></td>
<td>Use Windows threads mutex (Windows and MinGW only)</td>
</tr>
<tr>
<td><code>-Dmutex-manager=nothreads</code></td>
<td>Use dummy implementation (default if threading is disabled)</td>
</tr>
</table>
<p>Shared libraries are built by default. You can use the
<code>-DBUILD_SHARED_LIBS:BOOL=OFF</code> option to build
static libraries.</p>
<p>If you need to specify compiler executables that should be
used to build &XercesCName;, you can set the CC and CXX
environment variables when invoking
<code>cmake</code>. Similarly, if you need to specify
additional compiler or linker options, you can set the
CFLAGS, CXXFLAGS, and LDFLAGS environment variables. For
example:</p>
<source>CC=gcc-5.3 CXX=g++-5.3 CFLAGS=-O3 CXXFLAGS=-O3 cmake ...</source>
<note>
If building on Windows, the specific Visual Studio version
may be selected with some generators, and this may be run
from a normal command prompt. If using a generic generator
such as <code>Ninja</code>, then <code>cmake</code> should
be run from a Visual Studio command prompt, or in a
suitably configured environment, so that the correct
compiler will be detected.
</note>
<p>Once the configuration part is complete you can run the
build tool of choice. This may be done generically using
<code>cmake --build . [--config=Debug|Release]</code>.
Alternatively, a specific build tool, e.g. <code>make</code>,
<code>gmake</code>, <code>ninja</code> or
<code>msbuild</code> corresponding to the chosen generator
may be used directly. When invoked without a specific
target, it will build the &XercesCName; library, all examples
and all unit tests.</p>
<p>If you would like to run the automated test suite, run
<code>ctest [-V] [-C Debug|Release]</code>. This will run
all tests. Additional <jump
href="https://cmake.org/cmake/help/latest/manual/ctest.1.html">options</jump>
are available, such as running a subset of the tests and
running the tests in parallel. If any discrepancies in the
output are detected, the differences will be displayed if a
<code>diff</code> program is available.</p>
<p>Finally, install the library and examples. This may be
done generically using <code>cmake --build . --target
install</code>. Alternatively, a specific build tool may be
used, e.g. <code>make install</code>. To change the
installation directory, use the
<code>-DCMAKE_INSTALL_PREFIX=prefix</code> <code>cmake</code>
option.</p>
<p>Some platforms and configurations may require extra
<code>cmake</code> options. Run <code>cmake -LH</code> to
list the additional options, along with a short description
for each. For each of the selection categories mentioned
above, the help text will list the valid choices detected for
your platform. Run <code>cmake -LAH</code> for all the
additional advanced settings.</p>
<p>Several examples of configuring, building, testing and
installing with CMake using different platforms, generators,
and installation options are shown below:</p>
<table>
<tr>
<th>Platform</th>
<th>Generator</th>
<th>Example</th>
</tr>
<tr>
<td>Any</td>
<td>Ninja</td>
<td><code>mkdir build</code><br/>
<code>cd build</code><br/>
<code>cmake -G Ninja -DCMAKE_INSTALL_PREFIX=/opt/xerces-c -DCMAKE_BUILD_TYPE=Release -Dnetwork-accessor=curl /path/to/xerces-c/source</code><br/>
<code>ninja</code><br/>
<code>ctest -V -j 8</code><br/>
<code>ninja install</code></td>
</tr>
<tr>
<td>Unix</td>
<td>Unix Makefiles</td>
<td><code>mkdir build</code><br/>
<code>cd build</code><br/>
<code>cmake -G "Unix Makefiles" -DCMAKE_INSTALL_PREFIX=/opt/xerces-c -DCMAKE_BUILD_TYPE=Debug -Dmessage-loader=icu /path/to/xerces-c/source</code><br/>
<code>make -j8</code><br/>
<code>make test</code><br/>
<code>make install</code></td>
</tr>
<tr>
<td>Windows</td>
<td>msbuild with VS2015 x64</td>
<td><code>mkdir build</code><br/>
<code>cd build</code><br/>
<code>cmake -G "Visual Studio 14 2015 Win64" -DCMAKE_INSTALL_PREFIX=D:\libs &nbsp;&nbsp; \path\to\xerces-c\source</code><br/>
<code>cmake --build . --config Debug</code><br/>
<code>ctest -V -C Debug -j 4</code><br/>
<code>cmake --build . --config Debug --target install</code></td>
</tr>
</table>
<p/>
<note>
Note that different UNIX platforms use different system
environment variables for finding shared libraries. On Linux
and Solaris, the environment variable name is
<code>LD_LIBRARY_PATH</code>, on AIX it is
<code>LIBPATH</code>, on Mac OS X it is
<code>DYLD_FALLBACK_LIBRARY_PATH</code>, and on HP-UX it is
<code>SHLIB_PATH</code>.
</note>
<note>
Note that Windows is different from the UNIX platforms in
the way it finds shared libraries at run time. While UNIX
platforms may use the <code>LD_LIBRARY_PATH</code>
environment variable, Windows uses the <code>PATH</code>
environment variable if the library is not in the same
directory as the executable.
</note>
</s3>
<anchor name="UNIX"/>
<s3 title="Building on UNIX/Linux/Mac OS X/Cygwin/MinGW platforms">
<p>For building on UNIX and UNIX-like (GNU/Linux, Max OS X,
Cygwin, MinGW-MSYS) platforms &XercesCName; uses the
GNU automake-based build systems and requires that you
have <jump href="http://www.gnu.org/software/make/make.html">GNU
make</jump> installed. On some platforms GNU make is called gmake
instead of make.</p>
<p>As with all automake-based projects the build process is divided
into two parts: configuration and building. The configuration
part is performed using the <code>configure</code> script that
can be found in the <code>&XercesC3SrcInstallDir;</code> directory.
The build part is performed by invoking <code>make</code>.</p>
<p>Besides the standard <code>configure</code> options which
you can view by running <code>configure --help</code>,
&XercesCName; provides a number of project-specific options
that are worth mentioning. You can specify one option for
each category outlined below. If you do not specify anything
for a particular category then <code>configure</code> will
select the most appropriate default. At the end of its
execution <code>configure</code> prints the selected
values for each category.</p>
<p>Net Accessor (used to access network resources):</p>
<table>
<tr>
<th>Option</th>
<th>Description</th>
</tr>
<tr>
<td><code>--enable-netaccessor-curl</code></td>
<td>use the libcurl library</td>
</tr>
<tr>
<td><code>--enable-netaccessor-socket</code></td>
<td>use plain sockets</td>
</tr>
<tr>
<td><code>--enable-netaccessor-cfurl</code></td>
<td>use the CFURL API (only on Mac OS X)</td>
</tr>
<tr>
<td><code>--enable-netaccessor-winsock</code></td>
<td>use WinSock (only on Windows, Cygwin, MinGW)</td>
</tr>
<tr>
<td><code>--disable-network</code></td>
<td>disable network support</td>
</tr>
</table>
<p>Transcoder (used to convert between internal UTF-16 and other encodings):</p>
<table>
<tr>
<th>Option</th>
<th>Description</th>
</tr>
<tr>
<td><code>--enable-transcoder-gnuiconv</code></td>
<td>use the GNU iconv library</td>
</tr>
<tr>
<td><code>--enable-transcoder-iconv</code></td>
<td>use the iconv library</td>
</tr>
<tr>
<td><code>--enable-transcoder-icu</code></td>
<td>use the ICU library</td>
</tr>
<tr>
<td><code>--enable-transcoder-macosunicodeconverter</code></td>
<td>use Mac OS X APIs (only on Mac OS X)</td>
</tr>
<tr>
<td><code>--enable-transcoder-windows</code></td>
<td>use Windows APIs (only on Windows, Cygwin, MinGW)</td>
</tr>
</table>
<p>Message Loader (used to access diagnostics messages):</p>
<table>
<tr>
<th>Option</th>
<th>Description</th>
</tr>
<tr>
<td><code>--enable-msgloader-inmemory</code></td>
<td>store the messages in memory</td>
</tr>
<tr>
<td><code>--enable-msgloader-icu</code></td>
<td>store the messages using the ICU resource bundles</td>
</tr>
<tr>
<td><code>--enable-msgloader-iconv</code></td>
<td>store the messages in the iconv message catalog</td>
</tr>
</table>
<p>XMLCh type (UTF-16 character type):</p>
<table>
<tr>
<th>Option</th>
<th>Description</th>
</tr>
<tr>
<td><code>--enable-xmlch-char16_t</code></td>
<td>use <code>char16_t</code> (requires a C++11 compiler)</td>
</tr>
<tr>
<td><code>--enable-xmlch-uint16_t</code></td>
<td>use <code>uint16_t</code> from
<code>&lt;cstdint&gt;</code> or
<code>&lt;stdint.h&gt;</code>, or another unsigned
16-bit type such as <code>unsigned short</code> if
the standard types are unavailable</td>
</tr>
<tr>
<td><code>--enable-xmlch-wchar_t</code></td>
<td>use <code>wchar_t</code> (Windows only)</td>
</tr>
</table>
<p>Thread support is enabled by default and can be disabled with the
<code>--disable-threads</code> option. If disabled,
it will not be possible to select a mutex manager other than
<code>nothreads</code>. If enabled, one of the following
mutex managers may be selected:</p>
<table>
<tr>
<th>Option</th>
<th>Description</th>
</tr>
<tr>
<td><code>--enable-mutexmgr-standard</code></td>
<td>Use Standard C++ mutex (requires a C++11 compiler)</td>
</tr>
<tr>
<td><code>--enable-mutexmgr-posix</code></td>
<td>Use POSIX threads (pthreads) mutex (only on UNIX and Cygwin)</td>
</tr>
<tr>
<td><code>--enable-mutexmgr-windows</code></td>
<td>Use Windows threads mutex (Windows and MinGW only)</td>
</tr>
<tr>
<td><code>--enable-mutexmgr-nothreads</code></td>
<td>Use dummy implementation (default if threading is disabled)</td>
</tr>
</table>
<p>By default <code>configure</code> selects both shared and static
libraries. You can use the <code>--disable-shared</code> and
<code>--disable-static</code> options to avoid building the
version you don't need.</p>
<p>Finally, to make the build process cleaner the &XercesCName;
build system hides actual compiler commands being executed
by <code>make</code>. If you would like to see those then you
can specify the <code>--disable-pretty-make</code> option.</p>
<p>If you need to specify compiler executables that should be
used to build &XercesCName;, you can set the CC and CXX
variables when invoking <code>configure</code>. Similarly,
if you need to specify additional compiler or linker options,
you can set the CFLAGS, CXXFLAGS, and LDFLAGS variables.
For example:</p>
<source>./configure --disable-static CC=gcc-4.3 CXX=g++-4.3 CFLAGS=-O3 CXXFLAGS=-O3</source>
<p>Once the configuration part is complete you can run
<code>make</code> (or <code>gmake</code>). Running
<code>make</code> from the <code>&XercesC3SrcInstallDir;</code>
directory builds &XercesCName; library and examples. The
library is placed into the <code>src/.libs</code> directory. If
you like to build only the library, you can run make from
<code>&XercesC3SrcInstallDir;/src</code>.</p>
<p>If you would like to build the tests and run the
automated test suite, run <code>make check</code>
from the <code>&XercesC3SrcInstallDir;</code>
directory. The automated test suite required
Perl and the <code>diff</code> command.</p>
<p>Finally, to install the library and examples you can run
<code>make install</code> (or <code>gmake install</code>).
To change the installation directory, use the <code>--prefix</code>
<code>configure</code> option.</p>
<p>Some platforms and configurations require extra
<code>configure</code> and <code>make</code> options
which are shown in the following table.</p>
<table>
<tr>
<th>Platform</th>
<th>Compiler</th>
<th>Options</th>
</tr>
<tr>
<td>Solaris x86</td>
<td>Sun CC</td>
<td><code>./configure CXX=CC CC=cc</code></td>
</tr>
<tr>
<td>Solaris x86-64</td>
<td>Sun CC</td>
<td><code>./configure CXX=CC CC=cc CFLAGS=-xarch=amd64 CXXFLAGS=-xarch=amd64</code><br/>
(for newer Sun CC versions use -m64 instead of -xarch=amd64)</td>
</tr>
<tr>
<td>Solaris SPARC</td>
<td>Sun CC</td>
<td><code>./configure CXX=CC CC=cc</code></td>
</tr>
<tr>
<td>Solaris SPARCv9</td>
<td>Sun CC</td>
<td><code>./configure CXX=CC CC=cc CFLAGS=-xarch=v9 CXXFLAGS=-xarch=v9</code><br/>
(for newer Sun CC versions use -m64 instead of -xarch=v9)</td>
</tr>
<tr>
<td>AIX PowerPC</td>
<td>IBM XL C++</td>
<td><code>./configure CXX=xlC_r CC=xlc_r</code><br/>
<code>gmake libxerces_c_la_LDFLAGS=-qmkshrobj</code></td>
</tr>
<tr>
<td>AIX PowerPC-64</td>
<td>IBM XL C++</td>
<td><code>export OBJECT_MODE=64</code><br/>
<code>./configure CXX=xlC_r CC=xlc_r CXXFLAGS=-q64 CFLAGS=-q64</code><br/>
<code>gmake libxerces_c_la_LDFLAGS=-qmkshrobj</code></td>
</tr>
<tr>
<td>HP-UX IA-64-32</td>
<td>HP aCC</td>
<td><code>./configure CXX=aCC CC=aCC CFLAGS=-mt CXXFLAGS=-mt LDFLAGS=-mt</code></td>
</tr>
<tr>
<td>HP-UX IA-64</td>
<td>HP aCC</td>
<td><code>./configure CXX=aCC CC=aCC CFLAGS="-mt +DD64" CXXFLAGS="-mt +DD64" LDFLAGS="-mt +DD64"</code></td>
</tr>
<tr>
<td>Mac OS X x86-64</td>
<td>GCC</td>
<td><code>./configure CFLAGS="-arch x86_64" CXXFLAGS="-arch x86_64" </code></td>
</tr>
<tr>
<td>Mac OS X PowerPC-64</td>
<td>GCC</td>
<td><code>./configure CFLAGS="-arch ppc64" CXXFLAGS="-arch ppc64"</code></td>
</tr>
<tr>
<td>Mac OS X x86/PowerPC</td>
<td>GCC</td>
<td><code>./configure --disable-dependency-tracking CFLAGS="-arch i386 -arch ppc" CXXFLAGS="-arch i386 -arch ppc"</code></td>
</tr>
<tr>
<td>Mingw x86</td>
<td>GCC</td>
<td><code>./configure LDFLAGS=-no-undefined</code></td>
</tr>
<tr>
<td>Cygwin x86</td>
<td>GCC</td>
<td><code>./configure LDFLAGS=-no-undefined</code></td>
</tr>
</table>
<p/>
<note>
Note that different UNIX platforms use different system
environment variable for finding shared libraries. On Linux
and Solaris, the environment variable name is
<code>LD_LIBRARY_PATH</code>, on AIX it is
<code>LIBPATH</code>, on Mac OS X it is
<code>DYLD_LIBRARY_PATH</code>, and on HP-UX
it is <code>SHLIB_PATH</code>.
</note>
<note>
Note that Cygwin and MinGW are different from the UNIX platforms
in the way they find shared libraries at run time. While UNIX
platforms may use the <code>LD_LIBRARY_PATH</code> environment
variable, Cygwin and MinGW use the <code>PATH</code> environment
variable.
</note>
</s3>
</s2>
</s1>