3. Information for developers

This document is intended to explain some of the more useful things within the tree, and provide a standard for working on the code.

3.1. General stuff — common subdirectory

String handling

Use snprintf(). It’s even provided with a compatibility module if the target system doesn’t have it natively.

If you use snprintf() to load some value into a buffer, make sure you provide the format string. Don’t use user-provided format strings, since that’s an easy way to open yourself up to an exploit.

Don’t use strcat(). We have a neat wrapper for snprintf() called snprintfcat() that allows you to append to char * with a format string and all the usual string length checking of snprintf() routine.

Error reporting

Don’t call syslog() directly. Use upslog_with_errno() and upslogx(). They may write to the syslog, stderr, or both as appropriate. This means you don’t have to worry about whether you’re running in the background or not.

The upslog_with_errno() routine prints your message plus the string expansion of errno. The upslogx() just prints the message.

fatal_with_errno() and fatalx() work the same way, but they also exit(EXIT_FAILURE) afterwards. Don’t call exit() directly.

Debugging information

The upsdebug_with_errno(), upsdebugx(), upsdebug_hex() and upsdebug_ascii() routines use the global nut_debug_level, so you don’t have to mess around with printf()'s and if's yourself. Use them.

Memory allocation

xmalloc(), xcalloc(), xrealloc() and xstrdup() all check the results of the base calls before continuing, so you don’t have to. Don’t use the raw calls directly.

Config file parsing

The configuration parser, called parseconf, is now up to its fourth major version. It has multiple entry points, and can handle many different jobs. It’s usually used for parsing files, but it can also take input a line at a time or even a character at a time.

You must initialize a context buffer with pconf_init() before using any other parseconf function. pconf_encode() is the only exception, since it operates on a buffer you supply and is an auxiliary function.

Escaping special characters and quoting multiple-word elements is all handled by the state machine. Using the same code for all config files avoids code duplication.


this does not apply to drivers. Driver authors should use the upsdrv_makevartable() scheme to pick up values from ups.conf file. Drivers should not have their own config files.

Drivers may have their own data files, such as lists of hardware, mapping tables, or similar. The difference between a data file and a config file is that users should never be expected to edit a data file under normal circumstances. This technique might be used to add more hardware support to a driver without recompiling.

<time.h> vs. <sys/time.h>

This is already handled by autoconf, so just #include "timehead.h" and you will get the right headers on every system.

3.2. Device drivers — main.c

The device drivers use main.c as their core.

To write a new driver, you create a file with a series of support functions that will be called by main. These all have names that start with upsdrv_, and they will be called at different times by main depending on what needs to happen.

See the driver documentation for information on writing drivers, and also refer to the skeletal driver in skel.c.

3.3. Portability

Avoid things that will break on other systems. All the world is not an x86 Linux box.

C comments

There are still older systems out there that don’t do C++ style comments.

/* Comments look like this. */
// Not like this.

Variable declarations go on top

Newer versions of gcc allow you to declare a variable inside a function after code, somewhat like the way C++ operates, like this:

function do_stuff(void)

        int a;

        a = do_something_else();

While this will compile and run on these newer versions, it will fail miserably for anyone on an older system. That means you must not use it.

Note that gcc only warns about this with -pedantic flag, and clang with a -Weverything (possibly -Wextra) flag, which can be enabled by developers with configure --enable-warnings=... option values (and made fatal with configure --enable-Werror), to ensure non-regression of code quality. It was reported that clang-16 with such options does complain about non-portability to older C language revisions even if explicitly building for a newer revision.

Please note that for the purposes of legacy-compatible variable declarations (on top of their scopes), a NUT_UNUSED_VARIABLE(varname) counts as code and should be used just below the declarations. Initial assignments to variables (also as return values of methods) may generally happen as part of their declarations.

You can use scoping (e.g. do { ... } while (0);) where it makes sense to constrain visibility of temporary variables, such as in switch/case blocks.

Variable declaration in loop block syntax

Another feature that does not work on some compilers (e.g. conforming to "ANSI C"/C89/C90 standard) is initial variable declaration inside a for loop block, like this:

function do_stuff(void)
        /* This should declare "int i;" first, then use it in "for" loop: */
        for (int i = 0; i < INT_MAX; ++i) { ... }

        /* Additional loops cause also an error about re-declaring a variable: */
        for (int i = 10; i < 15; ++i) { ... }

Other hints


At this point NUT is expected to work correctly when built with a "strict" C99 (or rather GNU99 on many systems) or newer standard.

The NUT codebase may build in a mode without warnings made fatal on C89 (GNU89), but the emitted warnings indicate that those binaries may crash. By the end of 2021, NUT codebase has been revised to pass GNU and strict-C mode builds with C89 standard with the GCC toolkit (and on systems that do have the newer features in libraries, just hide them in standard headers); however CLANG toolkit is more restrictive about the C99+ syntax used. That said, some systems refuse to expose methods or types available in their system headers and binary libraries if strict-C mode is used alone, without extra system-specific defines to enable more than the baseline.

It was also seen that cross-builds (e.g. NUT for Windows using mingw on Linux) may be unable to define WIN32 and/or find symbols for linking when using a strict-C language standard.

The C support expects C11 or newer (not really configured or tested for older C98 or C03), modulo features that were deprecated in later language revisions (C++14 onwards) as highlighted by warnings from newer compilers.

Note also that the NUT codebase currently relies on certain features, such as the printf format modifiers for (s)size_t, use of long long, some nuances about structure/array initializers, variadic macros for debugging, etc. that a pedantic C90 mode compilation warns is not part of the standard but a GNU extension (and part of C99 and newer standard revisions). Many of the "offences" against the older standard actually come from system and third-party header files.

That said, the NUT CI farm does run non-regression builds with GNU C89 and "strict" C89 standard revisions and minimal passing warnings level, to ensure that codebase is and remains at least basically compliant. We try to cover a few distributions from early 2000’s for this, either in regular CI builds or one-off local builds for community members with a zoo of old systems.

If somebody in the community actually requires to build and run NUT on systems that old, where newer compilers are not available, pull requests to fix the offending coding issues in some way that does not break other use-cases are welcome.

3.4. Continuous Integration and Automated Builds

To ease and automate the build scenarios which were deemed important for quality assurance and non-regression checks of NUT, several solutions were introduced over time.

Build automation tools and scripts


This script was originally introduced (following ZeroMQ/ZProject example) to automate CI builds, by automating certain scenarios driven by exported environment variables to set particular configure options and make some targets (chosen by the BUILD_TYPE envvar). It can also be used locally to avoid much typing to re-run those scenarios during development.

Developers can directly use the scripts involved in CI builds to fix existing code on their workstations or to ensure support for new compilers and C standard revisions, e.g. save a local file like this to call the common script with pre-sets:

$ cat _fightwarn-gcc10-gnu17.sh
BUILD_TYPE=default-all-errors \
CFLAGS="-Wall -Wextra -Werror -pedantic -std=gnu17" \
CXXFLAGS="-Wall -Wextra -Werror -std=gnu++17" \
CC=gcc-10 CXX=g++-10 \

…and then execute it to prepare a workspace, after which you can go fixing bugs file-by-file running a make after each save to confirm your solutions and uncover the next issue to address :-)

Helpfully, the NUT CI farm build logs report the configuration used for each executed stage, so if some build combination fails — you can just scroll to the end of that section and copy-paste the way to reproduce an issue locally (on an OS similar to that build case).

Note that while spelling out sets of warnings can help in a quest to fix certain bugs during development (if only by removing noise from classes of warnings not relevant to the issue one is working on), there is a reasonable set of warnings which NUT codebase actively tries to be clean about (and checks in CI), detailed in the next section.

For the ci_build.sh usage like above, one can instead pass the setting via BUILD_WARNOPT=..., and require that all emitted warnings are fatal for their build, e.g.:

$ cat _fightwarn-clang9-gnu11.sh
BUILD_TYPE=default-all-errors \
CFLAGS="-std=gnu11" \
CXXFLAGS="-std=gnu++11" \
CC=clang-9 CXX=clang++-9 CPP=clang-cpp \

Finally, for refactoring effort geared particularly for fighting the warnings which exist in current codebase, the script contains some presets (which would evolve along with codebase quality improvements) as BUILD_TYPE=fightwarn-gcc, BUILD_TYPE=fightwarn-clang or plain BUILD_TYPE=fightwarn:

BUILD_TYPE=fightwarn-clang ./ci_build.sh

As a rule of thumb, new contributions must not emit any warnings when built in GNU99 mode with a minimal "difficulty" level of warnings. Technically they must survive the part of test matrix across the several platforms tested by NUT CI and marked in project settings as required to pass, to be accepted for a pull request merge.

Developers aiming to post successful pull requests to improve NUT can pass the --enable-warnings option to the configure script in local builds to see how that behaves and ensure that at least in some set-up their contribution is viable. Note that different compiler versions and vendors (gcc/clang/…), building against different OS and third-party dependencies, with different CPU architectures and different language specification revisions, might all complain about different issues — and catching this in as diverse range of set-ups as possible is why we have CI tests.

It can be beneficial for serial developers to set up a local BuildBot, Travis or a Jenkins instance with a matrix test job, to test their local git repository branches with whatever systems they have available.

While autoconf tries its best to provide portable shell code, sometimes there are builds of system shell that just fail under stress. If you are seeing random failures of ./configure script in different spots with the same inputs, try telling ./ci_build.sh to loop configuring until success (instead of quickly failing), and/or tell ./configure to use another shell at least for the system call-outs, with options like these:

SHELL=/bin/bash CONFIG_SHELL=/bin/bash CI_SHELL_IS_FLAKY=true \
Jenkins CI

Since mid-2021, the NUT CI farm is implemented by several virtual servers courteously provided by Fosshost and later by DigitalOcean.

These run various operating systems as build agents, and a Jenkins instance to orchestrate the builds of NUT branches and pull requests on those agents.

This is driven by Jenkinsfile-dynamatrix and a Jenkins Shared Library called jenkins-dynamatrix which prepares a matrix of builds across as many operating systems, bitnesses/architectures, compilers, make programs and C/C++ revisions as it can — based on the population of currently available build agents and capabilities which they expose as agent labels.

This hopefully means that people interested in NUT can contribute to the build farm (and ensure NUT is and remains compatible with their platform) by running a Jenkins Swarm agent with certain labels, which would dial into https://ci.networkupstools.org/ controller. Please contact the NUT maintainer if you want to participate in this manner.

The Jenkinsfile-dynamatrix recipe allows NUT CI farm to run different sets of build scenarios based on various conditions, such as the name of branch being built (or PR’ed against), changed files (e.g. C/C++ sources vs. just docs), and some build combinations may be not required to succeed.

For example, the main development branch and pull requests against it must cleanly pass all specified builds and tests on various platforms with the default level of warnings specified in the configure script. These are balanced to not run too many build scenarios overall, but just a quick and sufficiently representative set.

As another example, there is special handling for "fightwarn" pattern in the branch names to run many more builds with varying warning levels and more variants of intermediate language revisions, and so expose concerns deliberately missed by default warnings levels in "master" branch builds (the bar moves over time, as some classes of warnings become extinct from our codebase).

Further special handling for branches named like fightwarn.*89.* regex enables more intensive warning levels for a GNU89 build specifically (which are otherwise disabled as noisy yet not useful for supported C99+ builds), and is intended to help develop fixes for support of this older language revision, if anyone would dare.

Many of those unsuccessful build stages are precisely the focus of the "fightwarn" effort, and are currently marked as "may fail", so they end up as "UNSTABLE" (seen as orange bubbles in the Jenkins BlueOcean UI, or orange cells in the tabular list of stages in the legacy UI), rather than as "FAILURE" (red bubbles) for build scenarios that were not expected to fail and usually represent higher-priority problems that would block a PR.

Developers whose PR builds (or attempts to fix warnings) did not succeed in some cell of such build matrix, can look at the individual logs of that cell. Beside indication from the compiler about the failure, the end of log text includes the command which was executed by CI worker and can be reproduced locally by the developer, e.g.:

22:26:01  FINISHED with exit-code 2 cmd:  (
22:26:01  [ -x ./ci_build.sh ] || exit
22:26:01  eval BUILD_TYPE="default-alldrv" BUILD_WARNOPT="hard" \
    BUILD_WARNFATAL="yes" MAKE="make"  CC=gcc-10 CXX=g++-10 \
    CPP=cpp-10 CFLAGS='-std=gnu99 -m64' CXXFLAGS='-std=gnu++11 -m64' \
    LDFLAGS='-m64' ./ci_build.sh
22:26:01  )

or for autotools-driven scenarios (which prep, configure, build and test in separate stages — so for reproducing a failed build you should also look at its configuration step separately):

22:28:18  FINISHED with exit-code 0 cmd:  ( [ -x configure ] || exit; \
    eval  CC=clang-9 CXX=clang++-9 CPP=clang-cpp-9 CFLAGS='-std=c11 -m64' \
    CXXFLAGS='-std=c++11 -m64' LDFLAGS='-m64' time ./configure )

To re-run such scenario locally, you can copy the line from eval (but without the eval keyword itself) up to and including the executed script or tool, into your shell. Depending on locally available compilers, you may have to tweak the CC, CXX and CPP arguments; note that a CPP may be specified as /path/to/CC -E for GCC and CLANG based toolkits at least, if they lack a standalone preprocessor program (e.g. IntelCC).


While NUT recipes do not currently recognize a separate CXXCPP, it would follow similar semantics.

Some further details about the NUT CI farm workers are available in config-prereqs.txt and ci-farm-lxc-setup.txt documents.

AppVeyor CI

Primarily used for building NUT for Windows on Windows instances provided in the cloud — and so ensure non-regression as well as downloadable archives with binary installation prototype area, intended for enthusiastic testing (proper packaging to follow). NUT for Windows build-ability was re-introduced soon after NUT 2.8.0 release.

This relies on a few prerequisite packages and a common NUT configuration, as coded in the appveyor.yml file in the NUT codebase.


Primarily used for building NUT for MacOS on instances provided in the cloud, and so ensure non-regression across several Xcode releases.

This relies on a few prerequisite packages and a common NUT configuration, as coded in the .circleci/config.yml file in the NUT codebase.

Travis CI

See the .travis.yml file in project sources for a detailed list of third party dependencies and a large matrix of CFLAGS and compiler versions last known to work or to not (yet) work on operating systems available to that CI solution.


The cloud Travis CI offering became effectively defunct for open-source projects in mid-2021, so the .travis.yml file in NUT codebase is not actively maintained.

Local private deployments of Travis CI are possible, so if anybody does use it and has updated markup to share, they are welcome to post PRs.

The NUT project on GitHub has integration with Travis CI to test a large set of compiler and option combinations, covering different versions of gcc and clang, C standards, and requiring to pass builds at least in a mode without warnings (and checking the other cases where any warnings are made fatal).

Pre-set warning options

The options chosen into pre-sets that can be selected by configure script options are ones we use for different layers of CI tests.

Values to note include:

  • --enable-Werror(=yes/no) — make warnings fatal;
  • --enable-warnings(=.../no) — enable certain warning presets:

    • gcc-hard, clang-hard, gcc-medium, clang-medium, gcc-minimal, clang-minimal, all — actual definitions that are compiler-dependent (the latter just adds -Wall which may be relatively portable);
    • hard, medium or minimal — if current compiler is detected as CLANG or GCC, apply corresponding setting from above (or all otherwise);
    • gcc or clang — apply the set of options (regardless of detected compiler) with default "difficulty" hard-coded in configure script, to tweak as our codebase becomes cleaner;
    • yes/auto (also takes effect if --enable-warnings is requested without an =ARG part) — if current compiler is detected as CLANG or GCC, apply corresponding setting with default "difficulty" from above (or all otherwise).

Note that for backwards-compatibility reasons and to help filter out introduction of blatant errors, builds with compilers that claim GCC compatibility can enable a few easy warning presets by default. This can be avoided with an explicit argument to --disable-warnings (or --enable-warnings=no).

All levels of warnings pre-sets for GCC in particular do not enforce the -pedantic mode for builds with C89/C90/ANSI standard revision (as guesstimated by CFLAGS content), because nowadays it complains more about the system and third-party library headers, than about NUT codebase quality (and "our offenses" are mostly something not worth fixing in this era, such as the use of __func__ in debug commands). If there still are practical use-cases that require builds of NUT on pre-C99 compiler toolkits, pull requests are of course welcome — but the maintainer team does not intend to spend much time on that.

Hopefully this warnings pre-set mechanism is extensible enough if we would need to add more compilers and/or "difficulty levels" in the future.

Finally, note that such pre-set warnings can be mixed with options passed through CFLAGS or CXXFLAGS values to your local configure run, but it is up to your compiler how it interprets the resulting mix.

3.5. Integrated Development Environments (IDEs) and debugging NUT

Much of NUT has been coded using classic editors of developers' preference, like vi, nano, Midnight Commander mcedit, gedit/pluma, NotePad++ and tools like meld or WinMerge for file comparison and merge.

Modern IDEs however do offer benefits, specifically for live debugging sessions in a more convenient fashion than with command-line gdb directly. They also simplify writing AsciiDoc files with real-time rendering support.


Due to use of libtool wrappers in "autotools" driven projects, it may be tricky to attach the debugger (mixing the correct LD_LIBRARY_PATH or equivalent with a binary under a .libs subdirectory; on some platforms you may be better off copying shared objects to the directory with the binary being tested).

IDEs that were tested to work with NUT development and real-time debugger tracing include:

  • Sun NetBeans 8.2 on Solaris, Linux (including local and remote build and debug ability);
  • Apache NetBeans 17 on Windows with MSYS2 support (as MinGW toolkit);
  • Visual Studio Code (VSCode) on Windows with MSYS2 support.

Some supporting maintenance and development is doable with IntelliJ IDEA, making some things easier to do than with a simple Notepad, but it does not handle C/C++ development as such.

Take note that some IDEs can store their project data in the source root directory of a project (such as NUT codebase). While .gitignore rules can take care of not adding your local configuration into the SCM, these locations can be wiped by a careless git clean -fdX. You are advised to explore configuring your IDE to store project configurations outside the source codebase location, or to track such directories as nbproject or nb-cache as a separate Git repository (not necessarily a submodule of NUT nor really diligently tracked) to avoid such surprises.

IDE notes on Windows

General settings for builds on Windows

When working in a native Windows environment with MSYS2 (providing MinGW x64 among other things), you may need to ensure certain environment variables are set before you start the IDE (shortcuts and wrappers that start your console apply them via shell).


If you set such environment variables system-wide for your user profile (or wrap the IDE start-up by a script to set them), it may compromise your ability to use other MSYS2 profiles and/or other builds of these toolkits (packaged by e.g. Git for Windows or PERL for Windows projects) generally, or in the same IDE session, respectively. You may want to do this in a dedicated user account!

Examples below assume you installed MSYS2 into C:\msys64 (by default) and are using the "MinGW X64" profile for GCC builds (nuances may differ for 32-bit, CLANG, UCRT and other profile variants).

Also keep in mind that not all dependencies and tools involved in a fully-fledged NUT build are easily available or usable on Windows (e.g. the spell checker). See the config-prereqs.txt for better detailed package lists for different operating systems including Windows, and feel welcome to post pull requests with suggestions about new tool-chains that might fare better than those already tried and documented.

  • Make sure its tools are in the PATH:

    Control Panel ⇒ "Edit the system environment variables" ⇒ "Environment variables…" (button) ⇒ "Edit…" or create "New…" Path setting ("User variable" level suffices) ⇒

    • Make sure C:\msys64\mingw64\bin and C:\msys64\usr\bin are both there.
    • Depending on further installed toolkits, you may want to add C:\Program Files\Git\cmd or C:\Program Files\Microsoft VS Code\bin (preferably use deployment-dependent spellings without white-space like Progra~1 to err on the safe side of variable expansions later).
  • Make sure that MSYS2 (and tools which integrate with it) know its home:

    Open Environment variables window as above, and "Edit…" or create "New…" MSYS_HOME setting ⇒ Set to C:\msys64\mingw64\bin * Restart the IDE (if already running) for it to acknowledge the system configuration change.

Otherwise, NetBeans for example claims there is no shell for it to run make or open Terminal pane windows, and fails to start the built programs due to lack of DLL files they were linked against (such as libssl usually needed for any networked part of the codebase).

You might still have to fiddle with DLL files built in other directories of the NUT project, when preparing to debug certain programs, e.g. for dummy-ups testing you may need to:

:; cp ./clients/.libs/libupsclient-6.dll ./drivers/.libs/

To ensure builds with debug symbols, you may add CFLAGS and CXXFLAGS set to -g3 -gdwarf-2 or similar to configure options, or if that confuses the cross-build (it tends to assume those values are part of GCC path), you may have to hack them into your local copy of configure.ac, after the AM_INIT_AUTOMAKE([subdir-objects]) line:

CFLAGS="$CFLAGS -g3 -gdwarf-2"
CXXFLAGS="$CXXFLAGS -g3 -gdwarf-2"

…and re-run the ./autogen.sh script.

GDB on Windows

Examples below assume that whichever IDE you are using, the primary goal is to debug some issues with NUT on that platform.

This may require you to craft a configuration file for the GNU Debugger, e.g. C:\Users\abuild\.gdbinit for the examples below. One is not required however, and may be missing.

Another thing to keep in mind is that with libtool involved, the actual binary for testing would be in a .libs subdirectory and you may have some fun with ensuring that DLLs are found to start them — see the notes above.

NetBeans on Windows

When you install newer Apache NetBeans releases (14, 17 as of this writing), you may need to enable the use of "NetBeans 8.2 Plugin Portal" (check under Tools/Plugins/Settings) and install the "C/C++" plugin only available there at the moment. In turn, that older build of a plugin package may require that your system provides the unpack200(.exe) tool which was shipped with JDK11 or older (you may have to install that just to get the tool, or copy its binary from another system).

Under Tools/Options menu open the C/C++ tab and further its Build Tools sub-tab.


NetBeans allows you to easily define different Tool Collections, including those associated with a different build host (accessible over SSH and source/build paths optionally shared over NFS or similar technology, or copied over). This allows you to run the IDE on your desktop while debugging a build running on a server or embedded system.

Make sure you have a MinGW Tool Collection for the "localhost" build host with such settings as:

Option name

Sample value





Base Directory


C Compiler


C++ Compiler




Make Command


Debugger Command


In the Code Assistance sub-tab check that there are toolkit-specific and general include paths, e.g. both C and C++ Compiler settings might involve:





On top of that, C++ Compiler settings may include:




In the "Other" sub-tab, set default standards to C99 and C++11 to match common NUT codebase expectations.

Finally, open/create a "nut" project pointing to your git checkout workspace.

Next part of configuration regards build/debug configurations, which you can find on the toolbar or as File / Project Properties.

The main configuration for debugging a particular binary (and NUT has tons of those, good luck in case you want to debug several simultaneously) is in the Run and Debug categories. You may want to define different Configuration profiles to track the individual Run/Debug settings for different tested binaries, while the Build/Make settings would remain the same. Alternatively, you may set the Make category’s "Build Result" as the path to the binary you would test, and use ${OUTPUT_PATH} variable as its name in the "Run Command" (still likely need custom arguments) and "Symbol File" below.

When you investigate interactions of two or more programs, but only want to debug (step through) just one of them, you are advised to run each of the others from a dedicated terminal session, and just bump their debug verbosity.

  • In the Build category, set the Build Host (localhost) and Tool Collection (MinGW). In expert part of the settings, un-check "platform-independent" and revise that the TOOLS_PATH=C:\msys64\mingw64\bin while the UTILITIES_PATH=C:\msys64\usr\bin.
  • In the Pre-Build category likely keep the Working Directory as . and the Pre-Build First generally unchecked (so only enable it to reconfigure the project, which takes time and is not needed for every rebuild iteration), but you may still pre-set the Command line to something like the following (on one line):

    bash -c "rm -f configure Makefile; ./autogen.sh &&
        ./configure CC='${IDE_CC}' CXX='${IDE_CXX}'
            --with-all=auto --with-docs=skip"

    In some cases, NOT specifying the CC, CXX and the flags actually succeeds while passing their options fails the configuration ("Compiler can not create executables" etc.) probably due to path resolution issues between the native and MinGW environments.


    In practice, you may have an easier time using NUT ./ci_build.sh helper or running a more specific ./autogen.sh && ./configure ... spell similar to the above example or customized otherwise, in the MinGW x64 console window to actually configure a NUT source code setup, than to maintain one via the IDE. Running (re-)builds with the IDE (as you just edit non-recipe sources and iterate with a debugger) using externally configured Makefiles works fine.

  • In the Make category you may want to customize for parallelized builds on multi-CPU systems with something like:

    • Build Command: ${MAKE} -j 6 -f Makefile
    • Clean Command: ${MAKE} -f Makefile clean
  • In the Run category you should set the "Run Command" to point to your binary (note the .libs sub-directory, and see comments above regarding possibly needed copies of shared objects) and its arguments (all on one line), e.g.:

    C:\Users\abuild\Desktop\nut\drivers\.libs\usbhid-ups.exe -s ups -x port=auto
        -d1 -DDDDDD

    Other useful settings may be to keep "Build First" checked, and if the "Internal Terminal" does not work for you as the debugged program’s console — set the "Console Type" to "External Terminal" of type "Command Window". Unfortunately, NetBeans on Windows may have issues running terminal tabs unless CygWin is installed.

  • In the Debug category you should set the "Symbol File" to point to your tested binary (e.g. C:\Users\abuild\Desktop\nut\drivers\.libs\usbhid-ups.exe to match the "Run Command" example above) and specify "Follow Fork Mode" as "child" and "Detach On Fork" as "off". "Reverse Debugging" may be useful too in some situations. Finally, select your "Gdb Init File" if you have one, e.g. C:\Users\abuild\.gdbinit.
Microsoft VS Code

With this IDE you can benefit from numerous Extensions from its Marketplace, the ones found useful for NUT development and debugging include:

  • AsciiDoc (by asciidoctor)
  • EditorConfig for VS Code (by EditorConfig)
  • C/C++ (by Microsoft)
  • C/C++ Extension pack (by Microsoft)
  • Makefile tool (by Microsoft)
  • MSYS2/Cygwin/MinGW/Clang support (by okhlybov)
  • Native Debug (GDB, LLDB … Debugger support; by WebFreak)

Configurations are tracked locally in JSON files where you would need to add some entries. Examples below highlight the needed keys and values; your files may have others:

  • .vscode/launch.json (can create one via Run/Add Configuration… menu defines ways to launch the debug session for a program:

        "configurations": [
                "name": "CPPDBG GDB usbhid-ups",
                "type": "cppdbg",
                "request": "launch",
                "program": "C:\\Users\\abuild\\Desktop\\nut\\drivers\\.libs\\usbhid-ups.exe",
                "additionalSOLibSearchPath": "C:\\Users\\abuild\\Desktop\\nut\\.inst\\mingw64\\bin",
                "stopAtConnect": true,
                "args": ["-s", "ups", "-DDDDDD", "-d1", "-x", "port=auto"],
                "stopAtEntry": false,
                "cwd": "C:\\Users\\abuild\\Desktop\\nut",
                "environment": [],
                "externalConsole": false,
                "MIMode": "gdb",
                "miDebuggerPath": "C:\\msys64\\mingw64\\bin\\gdb.exe",
                "targetArchitecture": "x64",
                "setupCommands": [
                        "description": "Enable pretty-printing for gdb",
                        "text": "-enable-pretty-printing",
                        "ignoreFailures": true
                        "description": "Set Disassembly Flavor to Intel",
                        "text": "-gdb-set disassembly-flavor intel",
                        "ignoreFailures": true
                "preLaunchTask": "make usbhid-ups"
                // Alternately with LLDB (clang), the rest looks like above:
                "name": "CPPDBG LLDB usbhid-ups",
                "MIMode": "lldb",
                "miDebuggerPath": "C:\\msys64\\usr\\bin\\lldb.exe",
  • .vscode/tasks.json defines other tasks, such as the preLaunchTask mentioned above (assuming you have configured the build externally in the MinGW x64 terminal session):

        "tasks": [
                "type": "shell",
                "label": "make usbhid-ups",
                "command": "C:\\msys64\\usr\\bin\\make usbhid-ups",
                "options": {
                    "cwd": "${workspaceFolder}/drivers"
                "problemMatcher": [
                "group": {
                    "kind": "build",
                    "isDefault": true
  • .vscode/c_cpp_properties.json defines general compiler settings, e.g.:

        "configurations": [
                "name": "Win32",
                "includePath": [
                "defines": [
                "compilerPath": "C:\\msys64\\mingw64\\bin\\gcc.exe",
                "cStandard": "c99",
                "cppStandard": "c++11",
                "intelliSenseMode": "windows-gcc-x64",
                "configurationProvider": "ms-vscode.makefile-tools"
        "version": 4
IntelliJ IDEA

It is worth mentioning IntelliJ IDEA as another free (as of Community Edition) and popular IDE, however it is of limited use for NUT development.

Its ecosystem does feature a good AsciiDoc plugin, Python and of course the Java/Groovy support, so IDEA is helpful for maintenance of NUT documentation, helper scripts and CI recipes.

It lacks however C/C++ language support (allegedly a different product in the IntelliJ portfolio is dedicated to that), so for the core NUT project sources it is just a fancy text editor (with .editorconfig support) without syntax highlighting or codebase cross-reference aids, build/run/debug support, etc.

Still, it is possible to run builds and tests in embedded or external terminal session — so it is not worse than editing with legacy tools, and navigation or code-base-wide search is arguably easier.

3.6. Coding style

This is how we do things:

int open_subspace(char *ship, int privacy)
        if (!privacy)
                return insecure_channel(ship);

        if (!init_privacy(ship))
                fatal_with_errno("Can't open secure channel");

        return secure_channel(ship);

The basic idea is that we try to group things into functions, and then find ways to drop out of them when we can’t go any further. There’s another way to program this involving a big else chunk and a bunch of braces, and it can be hard to follow. You can read this from top to bottom and have a pretty good idea of what’s going on without having to track too much { } nesting and indenting.

We don’t really care for pretentiousVariableNamingSchemes, but you can probably get away with it in your own driver that we will never have to touch. If your function or variable names start pushing important code off the right margin of the screen, expect them to meet the byte chainsaw sooner or later.

All types defined with typedef should end in _t, because this is easier to read, and it enables tools (such as indent and emacs) to display the source code correctly.

Indenting with tabs vs. spaces

Another thing to notice is that the indenting happens with tabs instead of spaces. This lets everyone have their personal tab-width setting without inflicting much pain on other developers. If you use a space, then you’ve fixed the spacing in stone and have really annoyed half of the people out there.

Note that tabs apply only to indenting. Alignment of text after any non-tab character has appeared on the line must be done by spaces in order for it to remain at the same alignment when someone views tabs at a different widths.

One common example for this is multi-line if condition:

        if (something &&
            something_else) {

which may be written without mixing tabs and spaces to indent, as:

        if (something
        &&  something_else
        ) {

Another example is tables of definitions that are better aligned with (non-leading) spaces at least between names and values not too many characters wide; it still helps to align the columns with spaces at offsets divisible by 4 or 8 (consistently for the whole table):

#define SHORT_MACRO                         1   /* flag comment */
#define SOMETHING_WITH_A_VERY_LONG_NAME     255 /* flag comment */

If you write something that uses leading spaces, you may get away with it in a driver that’s relatively secluded. However, if we have to work on that code, expect it to get reformatted according to the above.

Patches to existing code that don’t conform to the coding style being used in that file will probably be dropped. If it’s something we really need, it will be grudgingly reformatted before being included.

When in doubt, have a look at Linus’s take on this topic in the Linux kernel — Documentation/CodingStyle. He’s done a far better job of explaining this.

Line breaks

It is better to have lines that are longer than 80 characters than to wrap lines in random places. This makes it easier to work with tools such as grep, and it also lets each developer choose their own window size and tab setting without being stuck to one particular choice.

Of course, this does not mean that lines should be made unnecessarily long when there is a better alternative (see the note on pretentiousVariableNamingSchemes above). Certainly there should not be more than one statement per line. Please do not use

if (condition) break;

but use the following:

if (condition) {


Earlier revisions of coding style might suggest avoiding braces if just one line is added as condition/loop/etc. handling code. Current approach is to welcome them even for single lines: on one hand, this confirms the intention that only this line is the conditional code; on another, this minimizes the context differences for later code comparisons, relocation, refactoring, etc.

Un-used variables and function arguments

Whenever a function needs to satisfy a particular API, it can end up taking arguments that are not used in practice (think a too-trivial signal handler). While some compilers offer the facility of decorations like __attribute__(unused), this proved not to be a portable solution. Also the abilities of newer C++ standard revisions are of no help to the vast range of existing systems that run NUT today and expect to be able to do so tomorrow (hence the required C99+ support noted above).

In NUT codebase we prefer to mark un-used variables explicitly in the body of the function (or an #ifdef branch of its code) using the NUT_UNUSED_VARIABLE(varname) as a routine call inside a function body, referring to the macro defined in common.h.

Please note that for the purposes of legacy-compatible variable declarations (on top of their scopes), NUT_UNUSED_VARIABLE(varname) counts as code and should happen below the declarations.

To display in a rough example:

        static void signal_X_handler(int signal_X) {
                /* We have explicitly got nothing to do if we catch signal X */

3.7. Miscellaneous coding style tools

NUT codebase includes an .editorconfig file which should be supported by most of the IDEs and text editors nowadays. Many support this format specification (at least partially) out of the box, possibly with some configuration toggle in the GUI. Others may need a plugin, see more at https://editorconfig.org/#pre-installed page. There are also command-line tools to verify and/or enforce compliance of source files to configuration.

You can go a long way towards converting your source code to the NUT coding style by piping it through the following command:

indent -kr -i8 -T FILE -l1000 -nhnl

This next command does a reasonable job of converting most C++ style comments (but not URLs and DOCTYPE strings):

sed 's#\(^\|[ \t]\)//[ \t]*\(.*\)[ \t]*#/* \2 */#'

Emacs users can adjust how tabs are displayed. For example, it is possible to set a tab stop to be 3 spaces, rather than the usual 8. (Note that in the saved file, one indentation level will still correspond to one tab stop; the difference is only how the file is rendered on screen). It is even possible to set this on a per-directory basis, by putting something like this into your .emacs file:

;; NUT style

(defun nut-c-mode ()
 "C mode with adjusted defaults for use with the NUT sources."
 (c-set-style "K&R")
 (setq c-basic-offset 3)  ;; 3 spaces C-indentation
 (setq tab-width 3))      ;; 3 spaces per tab

;; apply NUT style to all C source files in all subdirectories of nut/

(setq auto-mode-alist (cons '(".*/nut/.*\\.[ch]$". nut-c-mode)

Finishing touches

We like code that uses const and static liberally. If you don’t need to expose a function or global variable to the outside world, static is your friend. If nobody should edit the contents of some buffer that’s behind a pointer, const keeps them honest.

We always compile with -Wall, so things like const and static help you find implementation flaws. Functions that attempt to modify a constant or access something outside their scope will throw a warning or even fail to compile in some cases. This is what we want.

Switch case fall-through

While C standards allow to write switch statements to "fall through" from handling one case into another, modern compilers frown upon that practice and spew warnings which complicate detecting real bugs in the code (and also looking back at some of the cases written decades ago, it is not trivial to state whether the fall-through was intentional or really is a bug).

Compilers which detect such problem usually offer ways to decorate the code with comments or attributes to keep it quiet it in cases where the jump is intentional; also C++17 introduces special keywords for that in the standard. NUT aiming to be portable and independent of compilers as much as possible, prefers the arguably clearer and standards-based way of using goto into the next intended operation, even though it is a couple of lines away, e.g.:

int uppercase = 0;
switch (char_opt) {
        case 'U':
                uppercase = 1;
                goto fallthrough_case_u_option;
        case 'u':

In trivial cases, like falling through to default which just returns, it may be clearer and more maintainable (adding other option cases in the future) to just return same_result in the code block that would fall through otherwise and avoid goto statements altogether.


If you use a goto that jumps over long distances (see "Switch case fall-through" section above), expect us to drop it when our head stops spinning. It gives us flashbacks to the very old code we wrote. We’ve tried to clean up our act, and you should make the effort as well.

We’re not making a blanket statement about gotos, since everything probably has at least one good use. There are a few cases where a goto is more efficient than any other approach, but you probably won’t encounter them very often in this software.

Legacy code

There are parts of the source tree that do not yet conform to these specs. Part of this is due to the fact that the coding style has been evolving slightly over the course of the project. Some of the code you see in these directories is 5 years old, and things have gotten cleaner since then. Don’t worry — it’ll get cleaned up the next time something in the vicinity gets a visit.

Memory leak checking

We can’t say enough good things about valgrind. If you do anything with dynamic memory in your code, you need to use this. Just compile with gcc -g and start the program inside valgrind. Run it through the suspected area and then exit cleanly. valgrind will tell you if you’ve done anything dodgy like freeing regions twice, reading uninitialized memory, or if you’ve leaked memory anywhere.

For more information, refer to the Valgrind project.


The summary: please be kind to our eyes. There’s a lot of stuff in here, and many people have put a lot of time and energy to improve it.

3.8. Submitting patches

Current preference for suggesting changes is to open a pull request on GitHub for the https://github.com/networkupstools/nut/ project.

For some cases, small patches that arrive by mailing list in unified format (diff -u) as plain text attachments with no HTML and a brief summary at the top are easy to handle, but sadly also easy to overlook.

If a patch is sent to the nut-upsdev mailing list, it stands a better chance of being seen immediately. However, it is likely to be dropped if any issues cannot be resolved quickly. If your code might not work for others, or if it is a large change, your best bet is to submit a pull request or create an issue on GitHub.

The issue tracker allows us to track the patches over a longer period of time, and it is less likely that a patch will fall through the cracks. Posting a reminder to the developers (via the nut-upsdev list) about a patch on GitHub is fair game.

3.9. Patch cohesion

Patches should have some kind of unifying element. One patch set is one message, and it should all touch similar things. If you have to edit 6 files to add support for neutrino detection in UPS hardware, that’s fine.

However, sending one huge patch that does massive separate changes all over the tree is not recommended. That kind of patch has to be split up and evaluated separately, assuming the core developers care enough to do that instead of just dropping it.

If you have to make big changes in lots of places, send multiple patches — one per item.

3.10. The finishing touches: manual pages and device entry in HCL

If you change something that involves an argument to a program or configuration file parsing, the man page is probably now out of date. If you don’t update it, we have to, and we have enough to do as it is.

If you write a new driver, send in the man page when you send us the source code for your driver. Otherwise, we will be forced to write a skeletal man page that will probably miss many of the finer points of the driver and hardware.

The same remark goes for device entries: if you add support for new models, please remember to also complete the hardware compatibility list, present in data/driver.list.in. This will be used to generate both textual, static HTML and dynamic searchable HTML for the website.

Finally, don’t forget about fame and glory: if you added or substantially updated a driver, your copyright belongs in the heading comment (along with existing ones). For vendor backed (or sponsored) contributions we welcome an entry in the docs/acknowledgements.txt file as well, to track and know the industry players who help make NUT better and more useful.

It is nice to update the NEWS file for significant development to be seen as part of next release, as well as to update the UPGRADING file for potentially breaking changes and similar heads-up notes for third-party teams (distribution packagers, clients and bindings, etc.)

3.11. Source code management

We currently use a Git repository hosted at GitHub to track changes to the NUT source code. This allows you to clone the repository (or fork, in GitHub parlance), make changes, and post them online for peer review prior to integration.

To obtain permission to commit directly to the common upstream NUT repository, you must be prepared to spend a fair amount of time contributing to the NUT codebase. Most developers will be well served by committing to their own forked Git repository (preferably in a uniquely named branch for each new contribution), and having the NUT team merge their changes using pull requests.

Git offers a little more flexibility than the svn update command. You may fetch other developers' changes into your repository, but hold off on actually combining them with your branch until you have compared the two branches (for instance, with gitk --all). Git also allows you to accumulate more than one commit worth of changes before pushing to another repository. This allows development to continue without a constant network connection.

For a quick change to a file in the Git working copy, you can use git diff to generate a patch to send to the nut-upsdev mailing list. If you have more extensive changes, you can use git format-patch on a complete commit or branch, and send the resulting series of patches to the list.

If you use GitHub’s web-based editor to make changes, it tends to create lots of small commits, one per change per file. Unless there is reason to keep the intermediate history, we will probably collapse (or "squash" in Git parlance) the entire branch into one commit with a git rebase -i before merging.

The GitSvnCrashCourse wiki page has some useful information for long-time users of Subversion.

Git access

Anonymous Git checkouts are possible:

git clone git://github.com/networkupstools/nut.git


git clone https://github.com/networkupstools/nut.git

if it is necessary to get around a pesky firewall that blocks the native Git protocol.

For a quicker checkout (when you don’t need the entire repository history), you can limit the depth of the clone:

git clone --depth 1 git://github.com/networkupstools/nut.git

Mercurial (hg) access

There are those who prefer the simplicity and self-consistency of the Mercurial SCM client over the hodgepodge of unique commands which make up Git. Rather than debate the merits of each system, we will gently guide you towards the hg-git project which would theoretically be a transparent bridge between the central Git repository, and your local Mercurial working copy.

Other tools for hg/git interoperability are sure to exist. We would welcome any feedback about this process on the nut-upsdev mailing list.

Subversion (SVN) access

If you prefer to check out the NUT source code using an SVN client, GitHub has a SVN interface to Git repositories hosted on their servers. You can fork a copy of the NUT repository and commit to your fork with SVN.

Be aware that the examples in the GitHub blog post might result in a checkout that includes all of the current branches, as well as the trunk. You are most likely interested in a command line similar to the following:

svn co https://github.com/networkupstools/nut/trunk nut-trunk-svn

3.12. Ignoring generated files

The NUT repository generally only holds files which are not generated from other files. This prevents spurious differences from being recorded in the repository history.

If you add a driver, it is recommended that you add the driver executable name to the .gitignore file in that directory. Similarly, files generated from *.in and *.am source templates should be ignored as well. We try to include a number of generated files in the tarball releases with make dist hooks in order to minimize the number of dependencies for end users, but the assumption is that a developer can install the packages needed to regenerate those files.

3.13. Commit message formatting

From the git commit man page:

Though not required, it’s a good idea to begin the commit message with a single short (less than 50 character) line summarizing the change, followed by a blank line and then a more thorough description. The text up to the first blank line in a commit message is treated as the commit title, and that title is used throughout git.

If your commit is just a change to one component, such as the HCL, upsd or a specific driver, prefix your commit message in a way that matches similar commits. This helps when searching the repository or tracking down a regression.

Referring to previous commits can be tricky. If you are referring to the immediate parent of a given commit, it suffices to say "the previous commit". (Are you correcting a typo in the previous commit? If you haven’t pushed yet, consider using the git commit --amend command instead of creating a new commit.) For other commits, even though tools like gitk and GitHub’s repository viewers recognize Git hashes and create links automatically, it is best to add some context such as the commit title or a date.

You may notice that some older commits have [[SVN:####]] tags and Fossil-ID footers. These were lifted from the old SVN commit messages using reposurgeon, and should not be used as a guide for future commits.

3.14. Commit sign-off

Please also note that since 2023 we explicitly ask for contributions to be "Signed Off" according to "Developer Certificate of Origin" as represented in the LICENSE-DCO file in the root of NUT source tree (verbatim copy of Version 1.1 of DCO published at https://developercertificate.org/ web site). This is exactly the same one created and used by the Linux kernel developers.

This is a developer’s certification that he or she has the right to submit the patch for inclusion into the project. Simply submitting a contribution implies this agreement, however, please include a "Signed-off-by" tag in every patch (this tag is a conventional way to confirm that you agree to the DCO). In other words, this tag certifies that committer has the rights to submit this work under the same license as the project and agrees to the terms of a Developer Certificate of Origin.

Note that while git commit hook tricks are available to automatically sign off all commits, these signatures are intended to be a conscious (legally meaningful) act — hence they are not automated in git core with an easy configuration option.

For more details see:

You are also encouraged to set up a PGP key, make its public part known, and use it to sign your git commits (in addition to the Signed-Off-By tag) by also passing a -S option or calling git config commit.gpgsign true once. Numerous public articles can walk you through this ordeal, including:

3.15. Repository etiquette and quality assurance

For developers who have commit access to the common upstream NUT repository: Please keep the Git "master" branch in working condition at all times. The "master" branch may be used to generate daily tarballs, it provides the baseline for new contributions, and occasionally is tagged for a new release. It should not contain broken code. If you need to commit incremental changes that leave the system in a broken state, please do so in a separate branch and merge the changes back into "master" once they are complete.

To help keep the codebase ever-green, we run a number of CI tests and builds in various conditions, including older compilers, different C/C++ standard revisions, and an assortment of operating systems; a section below elaborates on this in more detail.

You are encouraged to use git rebase -i on your private Git branches to separate your changes into logical changes.

From there, you can generate patches for the issue tracker, or the nut-upsdev mailing list.

Note that once you rebase a branch, anyone else who has a copy of this branch will need to rebase on top of your rebased branch. Obviously, this hinders collaboration. In this case, we recommend that you rebase only in your private repository, and push when things are ready for discussion. Merging instead of rebasing will help with collaboration, but please do not turn the repository history into a pile of spaghetti by merging unnecessarily. (Test merges can be done on integration branches, which can be discarded if the merge is trivial.) Be sure that your commit messages are descriptive when merging.

If you haven’t created a commit out of your local changes yet, and you want to fetch the latest code, you can also use git stash before pulling, then git stash pop to apply your saved changes.

Here is an example workflow:

        git clone -o central git://github.com/networkupstools/nut.git

        cd nut
        git remote add -f username git://github.com/username/nut.git

        git checkout master
        git branch my-new-feature
        git checkout my-new-feature

        # Hack away

        git add changed-file.c
        git commit -s

        # Fix a typo in a file or commit message:

        git commit -s -a --amend

        # Someone committed something to the central repository. Fetch it.

        git fetch central
        git rebase central/master

        # Publish your branch to your GitHub repository:

        git push username my-new-feature

If you are new to Git, but are familiar with SVN, some of the following links may be of use:

3.16. Building the Code

For a developer, the NUT build process starts with ./autogen.sh.

This script generates the ./configure script that end users typically invoke to build NUT. If you are making a number of changes to the NUT source tree, configuring with the --enable-maintainer-mode flag will ensure that after you change a Makefile.am, nearby Makefile.in and Makefile get regenerated. At a minimum, you will need at least:

  • autoconf
  • automake
  • libtool
  • Python
  • Perl


See the config-prereqs.txt for better detailed package lists for different operating systems.

See ci_build.sh for automating many practical scenarios, for easier iterations.

It is optional, but highly recommended, to have Python 2.x or 3.x, and Perl, to generate some files included into the configure script whose presence is checked by autotools when it is generated. Neutered files can be just "touched" to pass the autogen.sh if these interpreters are not available, and effectively skip those parts of the build later on — autogen.sh will then advise which special environment variables to export in your situation and re-run it.

Even if you do not use your distribution’s packages of NUT, installing the distribution’s list of build dependencies for NUT can reduce the amount of trial-and-error when installing dependencies. For instance, in Debian, you can run apt-get build-dep nut to install all of the auto* tools as well as any development libraries and headers.

After running ./autogen.sh, you can pass your local configuration options to ./configure and run make from the top-level directory. To avoid the need for root privileges when testing new NUT code, you may wish to use --prefix=$HOME/local/nut --with-statepath=/tmp. You can also keep compilation times down by only building the driver which you are currently working on: --with-drivers=driver1,dummy-ups.

Before pushing your commits upstream, please run make distcheck-light. This checks that the Makefiles are not broken, that all the relevant files are distributed, and that there are no compilation or installation errors. Note that unless you specifically pass --with-doc=skip to configure, this requires all of the dependencies necessary to build the documentation to be locally installed on your system, including asciidoc, a2x, xsltproc, dblatex and any additional XSL stylesheets.

Running make distcheck-light is especially important if you have added or removed files, or updated configure.ac or some Makefile.am file. Remember: simply adding a file to Git does not mean it will be distributed. To distribute a file, you must update the corresponding Makefile.am with EXTRA_DIST entry and possibly other recipe handling.

There is also make distcheck, which runs an even stricter set of tests than make distcheck-light, but will not work unless you have all of the optional third-party libraries and features installed.

Finally note, that since 2017 the GitHub upstream project is monitored by Travis CI (in addition to earlier multi-platform buildbots which occasionally do not work), replaced since 2021 by a dedicated NUT CI farm. This means that if your posted improvements are based on current NUT "master" branch, the resulting pull request should get tested for a number of scenarios automatically. If your code adds a substantial feature, consider extending the Jenkinsfile-dynamatrix and/or ci_build.sh scripts in the workspace root to add another BUILD_TYPE to the matrix of tests run in parallel.