Requirements:
- GNU/Linux / Unix / macOS / BSD with bash compatible shell
- all common system progs (grep, cat, etc...)
- all common build requirements (libc, libstdc++, g++, GNU make, etc...)
- optionally CMake
or
- for 32-bit target: MinGW32 UCRT runtime for MS Windows, configured with MinGW bin directories added in
PATH
variable (typicallyC:\mingw32\bin;
), with any BASH emulator (e.g. the one which comes with git for windows) - for 64-bit target: MinGW64 UCRT runtime for MS Windows, configured with MinGW bin directories added in
PATH
variable (typicallyC:\mingw64\bin;
), with any BASH emulator (e.g. the one which comes with git for windows) or as an alternative MSYS2 environment with GNU Make and GCC installed.
C++17 standard is required, compilation is tested with GCC 14.2.0. It will not work with GCC 10.x and older. C++17 is required since v1.21.0, older versions required C++14 and can be built with GCC 5.x.
Use git
to clone the github repository and all git-submodules:
git clone --recursive -j8 https://github.com/z00m128/sjasmplus.git
If you already have cloned repository without submodules, you can update submodules only by:
git submodule update --init --recursive
If you are not using git
, you can download and extract archive from releases, the manually added .tar.xz
one (the automatically github provided zip files are missing the submodules sources and will fail to build). Or you can download the github provided archives, but then you need to download each submodule individually and extract them to correct folders inside sjasmplus folder.
Go to folder with sjasmplus project. Edit install path PREFIX
in file Makefile
according your preferences (default /usr/local
). Run following commands:
make clean
make
Then run as root or use sudo:
make install
Check cizo2000's github gentoo-overlay/dev-util/sjasmplus for ebuild files.
Recently (2022) it seems leo72 started providing AUR for z00m's fork of sjasmplus: aur.archlinux.org/packages/sjasmplus-z00m128 (AFAIK there's no connection with any maintainer of this project, but hopefully this will continue and provide good experience to AUR users, thanks leo72).
Go to folder with sjasmplus project. Edit install path PREFIX
in file Makefile.win
according your preferences (default c:\mingw32\usr\local\bin
). Remove -static
parameter in CFLAGS
if you don't need standalone Windows executable (binary is MinGW dependant then, but it's smaller). Run following commands:
mingw32-make -f Makefile.win clean
mingw32-make -f Makefile.win
mingw32-make -f Makefile.win install
Go to folder with sjasmplus project and run following set of commands:
cmake -DCMAKE_BUILD_TYPE=Release -S . -B build
cmake --build build
cmake --install build
The project will use internal copy of Lua 5.4.4 by default, but you can use system Lua 5.4 by using -DSYSTEM_LUA=ON
.
For disabling of LUA scripting support add -DENABLE_LUA=OFF
option:
cmake -DENABLE_LUA=OFF -DCMAKE_BUILD_TYPE=Release -S . -B build
Binary sjasmplus file will be placed in /usr/bin
by default.
To change install directory prefix add -DCMAKE_INSTALL_PREFIX:PATH
option with specified prefix:
cmake -DCMAKE_INSTALL_PREFIX:PATH=/usr/local -DCMAKE_BUILD_TYPE=Release -S . -B build
The CMake can generate VS project files, and sources should compile with VS compiler, since v1.14.1 the VS built executable should mostly work as well as MinGW, although there are still corner-case bugs which need to be fixed, see issues #77 and #78 for details/progress and to help.
The official binary "exe" is built with MinGW toolchain and that's recommended option.
There are few extra recommendations if you want to join sjasmplus development and use all the features:
- work with git clone of repository ideally, so you can easily update to latest source base, track your changes or review changes of other developers
- within the git repository init and update the submodules too, this will clone the UnitTest++ and LuaBridge2.8 repository which are required to build sjasmplus and unit tests (
git submodule init
,git submodule update
) - linux + gcc + Makefile is the config of Ped7g, having access to the same config to re-create his workflow locally may be of help (when troubleshooting some issue or comparing results with different platform)
- but having different local configuration would be very helpful to keep the source base cross-platform and in good shape
- you can check
.cirrus.yml
file and accompanying scripts/batch-files inContinuousIntegration
folder to see how different environments and different build tasks are prepared and executed. If you are not familiar with CI setup and configuration yet, you should take at least a brief glimpse on it, even if you want just to contribute small patch to the sjasmplus, because any pull request will be scrutinized by the CI build system automatically.
My (Ped7g) local workflow is:
- KDevelop IDE set up to use CMake import of project to build+install debug binary in my
~/.local/bin
folder, which is the first binary found by my search path (to build + install the sjasmplus I use Project/Install Selection [Shift+F8]). - with new debug binary locally installed, I run in terminal
ContinuousIntegration/test_folder_tests.sh
to verify if it works (adding new tests exercising the feature/change which I am working at). Sometimes I limit the scope of tests being run by adding argument to the script like "module" to run onlytests/modules/*
tests. - when I'm finished with a change, and I'm preparing commit, I use
make clean && make
to verify the release can be built too, andmake DEBUG=1 clean && make DEBUG=1 coverage
to run full test suite, C++ unit tests and create coverage report (text files in build directory) so I can manually check if everything did change as expected. - after this review I commit + push the changes to the repository, where the Cirrus CI setup will try to replicate my local experience in different environments of the CI, if some of these fails, I check the Cirrus logs to see why build failed, and work on the fix commit.
- when working on small/medium change where I'm sure I can get main repository into "green" state whenever I push changes to github, I tend to work directly on
master
branch, but when experimenting with things with unclear outcome, I use experimental branches, later merging them intomaster
when they are ready (or even my fork of the main repository, if the experiment is really just experiment, and I am not even sure if it will ever land in the main repository).
Enjoy!