This guide covers the build instructions for the stand-alone web server. See Embedding.md for information on extending an existing C or C++ application. A brief overview of the source code files can be found in Embedding.md as well.
Where to get the source code?
The latest development version can be found at https://github.com/civetweb/civetweb
Tested and released versions can be found at https://github.com/civetweb/civetweb/releases
Building for Windows
Using Visual Studio
Open the VS/civetweb.sln in Visual Studio. To include SSL support, you may have to add an extra library for the cryptography support. You might wish to use yaSSL. However, it is GPL licensed or uses a commercial license. See yaSSL.md for more information. Alternatively, you might wish to use OpenSSL. See OpenSSL.md for more information.
Using MinGW-w64 or TDM-GCC
In the start menu locate and run the “Run terminal” batch file. For TDM-GCC this is named “MinGW Command Prompt”. Navigate to the civetweb sources directory and run:
Using Qt Creator
Open the Qt Designer project in the Qt folder
Except for the components in the
third_party folder (e.g., Lua and Duktape), CivetWeb can also be built with CMake.
CMake can be used for all supported operating systems.
Building for Linux, BSD, and OSX
Get a list of all supported make option
make build make WITH_ALL=1
Compile the code. Using the option “WITH_ALL=1” enables all optional features.
Install on the system, Linux only.
make lib WITH_IPV6=1 make clean slib WITH_LUA=1 WITH_WEBSOCKET=1
Build the static and shared libraries. The additional make options configure the library just as it would the application.
The slib option should be done on a separate clean build as position independent code (PIC) is required for it. Trying to run it after building the static library or the server will result in a link error.
Clean up files generated during the build
Setting build options
Make options can be set on the command line with the make command like so.
make build WITH_LUA=1
||build with Lua support|
||with IPV6 support|
||build with web socket support|
||build with unix domain socket support|
||build with support for server statistics|
||include experimental features (version depending)|
||Include all of the above features|
||build with GDB debug support|
||build libraries with c++ classes|
||use ‘file’ as the config file|
||use ‘file’ as the backup config file|
||place to install initial web pages|
||default document root|
||listening ports override when installing|
||use versioned SSL library|
||system versioned CRYPTO library|
||sets the install directory|
||method to insert compile flags|
Note that the WITH_* options used for make are not identical to the preprocessor defines in the source code - usually USE_* is used there.
To change the target destination pass the
PREFIX option to the command
make install (not
make build). Example usage:
$ make build $ make -n install PREFIX=/opt/civetweb
-n corresponds to the
--dry-run option (it does not make any changes): You can see where
make install would install. Example output of the above command:
$ make -n install PREFIX=/opt/civetweb install -d -m 755 "/opt/civetweb/share/doc/civetweb" install -m 644 resources/itworks.html /opt/civetweb/share/doc/civetweb/index.html install -m 644 resources/civetweb_64x64.png /opt/civetweb/share/doc/civetweb/ install -d -m 755 "/opt/civetweb/etc" install -m 644 resources/civetweb.conf "/opt/civetweb/etc/" sed -i 's#^document_root.*$#document_root /opt/civetweb/share/doc/civetweb#' "/opt/civetweb/etc/civetweb.conf" sed -i 's#^listening_ports.*$#listening_ports 8080#' "/opt/civetweb/etc/civetweb.conf" install -d -m 755 "/opt/civetweb/share/doc/civetweb" install -m 644 *.md "/opt/civetweb/share/doc/civetweb" install -d -m 755 "/opt/civetweb/bin" install -m 755 civetweb "/opt/civetweb/bin/"
If the output looks good: Just remove the
-n option to actually install the software on your system.
Setting compile flags
Compile flags can be set using the COPT make option like so.
make build COPT="-DNDEBUG -DNO_CGI"
||strip off all debug code|
||build debug version (very noisy)|
||do not use atomic functions, use locks instead|
||disable caching functionality|
||disable CGI support|
||do not serve files from a directory|
||completely disable filesystems usage (requires NO_FILES)|
||disable nonce check for HTTP digest authentication|
||send all mg_response_header_* immediately instead of buffering until the mg_response_header_send call|
||disable SSL functionality|
||link against system libssl library|
||do not set a name for pthread|
||enable Application-Level-Protocol-Negotiation, required for HTTP2|
||enable HTTP2 support (experimental, not reccomended for production)|
||enable IPv6 support|
||enable Lua support|
||enable server statistics support|
||define stack size instead of using system default|
||enable websocket support|
||enable unix domain socket support|
||enable on-the-fly compression of files (using zlib)|
||include experimental interfaces|
||include obsolete interfaces (candidates for deletion)|
||disables large files (Lua only)|
||do not initialize libcrypto|
||Use OpenSSL V1.0.x interface|
||Use OpenSSL V1.1.x interface|
||Use OpenSSL V3.x interface|
||Use MbedTLS (cannot be combined with OPENSSL_API_*)|
||define as a string to be used as build id instead of DATE|
make is used (with this Makefile), you should not pass the
USE_<feature> flags using
COPT, but use the
WITH_<feature> syntax above, since additional features may also use additional source code files.
Take total control with CC, COPT and TARGET_OS as make options. TARGET_OS is used to determine some compile details as will as code function. TARGET_OS values should be be one found in resources/Makefile.in-os.
make CC=arm-none-linux-gnueabi-gcc COPT="-march=armv7-a -mfpu=vfp -mfloat-abi=softfp" TARGET_OS=FROG
Cocoa DMG Packaging (OSX Only)
Use the alternate Makefile.osx to do the build. The entire build has to be done using Makefile.osx because additional compile and link options are required. This Makefile has all the same options as the other one plus one additional package rule.
make -f Makefile.osx package
Building with Buildroot
Buildroot is a tool for creating cross compiled file systems. Including Civetweb in buildroot is fairly easy. There is even support for various build options.
- First, check if it already there.
- In buildroot, make menuconfig
- Package Selection for the target —>
- Networking applications —>
- If not there, just add it
- copy Config.in and civetweb.mk from Civetweb’s contrib/buildroot/ to Buildroot’s package/civetweb/ directory.
- In Buildroot’s *package/Config.in, insert the following line in were you will know how to find it in the menu.
Building on Android
This is a small guide to help you run civetweb on Android, originally tested on the HTC Wildfire. Note: You do not need root access to run civetweb on Android.
- Download the source from the Downloads page.
- Download the Android NDK from http://developer.android.com/tools/sdk/ndk/index.html
/path-to-ndk/ndk-build -C /path-to-civetweb/resourcesThat should generate civetweb/lib/armeabi/civetweb
- Using the adb tool (you need to have Android SDK installed for that),
push the generated civetweb binary to
/data/localfolder on device.
- From adb shell, navigate to
- To test if the server is running fine, visit your web-browser and
http://127.0.0.1:8080You should see the
Index of /page.
jnistands for Java Native Interface. Read up on Android NDK if you want to know how to interact with the native C functions of civetweb in Android Java applications.