man/bear.1.md
% BEAR(1) Bear User Manuals % László Nagy % July 3, 2026
<!-- to generate the final `bear.1` file, run `pandoc -s -t man bear.1.md -o bear.1` -->Bear - a tool to generate compilation database for Clang tooling.
bear [OPTIONS] [--] [BUILD_COMMAND...]
bear intercept [OPTIONS] [--] BUILD_COMMAND...
bear semantic [OPTIONS]
Bear is a tool that generates a JSON compilation database for Clang tooling by intercepting command executions during the build process. The JSON compilation database is used in the Clang project to provide information about how individual compilation units were processed, enabling tools like clang-tidy, clangd, and other Clang-based analysis tools to understand your project's build configuration.
Bear operates by intercepting system calls during the build process to capture compilation commands. It supports two main interception methods: dynamic library preloading (on Unix-like systems) and wrapper executables (cross-platform). The captured commands are then filtered through semantic analysis to identify actual compiler invocations and generate the final compilation database.
Bear can operate in three modes:
-c, --config FILE : Specify a configuration file path. The configuration file controls output formatting, compiler recognition, source filtering, and duplicate handling.
-o, --output FILE
: Specify the output file path (default: compile_commands.json). The output is a JSON compilation database.
-a, --append
: Append results to an existing output file instead of overwriting it. This allows incremental updates to the compilation database. New entries are placed before the existing ones, so when a source file is rebuilt its newest invocation survives duplicate filtering and replaces the stale entry (see the duplicates section).
-h, --help : Print help information.
-V, --version : Print version information.
Calling bear without commands will execute the combined mode, and will intercept the compiler calls and generate a compilation database as output.
Intercepts command execution events during the build process and saves them to an events file for later processing.
bear intercept [OPTIONS] [--] BUILD_COMMAND...
Processes previously captured events to generate a compilation database through semantic analysis.
bear semantic [OPTIONS]
Bear generates a JSON compilation database conforming to the Clang JSON Compilation Database specification. The output is a JSON array of compilation entry objects.
Each compilation database entry contains the following fields:
directory : The working directory of the compilation (absolute path)
file : The main translation unit source file (absolute path)
arguments : The compilation command as an array of strings (preferred format)
command : The compilation command as a single shell-escaped string (alternative to arguments)
output : The output file produced by compilation (optional, absolute path)
The output format can be controlled through the configuration file:
Bear generates entries where all paths are absolute by default, and uses the arguments field instead of command to avoid shell escaping issues.
Bear uses a YAML configuration file to control its behavior. The configuration file follows a structured schema with several main sections.
schema: "4.1"
intercept:
mode: wrapper
compilers:
- path: /usr/bin/cc
as: gcc
- path: /usr/local/bin/gcc
ignore: true
sources:
directories:
- path: /project/tests
action: exclude
duplicates:
match_on:
- file
- arguments
format:
paths:
directory: canonical
file: canonical
entries:
use_array_format: true
include_output_field: true
arguments:
from_response_files: false
from_environment: true
This example configuration file:
sets the interception mode to wrapper,
hints the /usr/bin/cc to be the main compiler in this project, which is the GNU compiler,
hints to ignore the /usr/local/bin/gcc compilers from the project,
instructs to ignore files from /project/tests,
instructs to detect duplicates based on the file and arguments fields of the output file,
instructs to format the output to use canonical path for the file and directory fields of the output file,
instructs to use the arguments over the command field in the output file,
instructs to include the output field in the output file.
The configuration file uses schema version 4.1 and has the following structure:
Controls the command interception method:
preload (Unix) or wrapper (cross-platform)Contains hints about what compiler needs to be recognized and what that compiler is.
gcc, clang, flang, intel-fortran, cray-fortran, cuda, msvc, clang-cl, intel_cc, nvidia-hpc, armclang, ibm_xl, vala.The generic compiler names cc and c++ default to GCC semantics. On platforms where these map to a different compiler (macOS, FreeBSD, OpenBSD), use the as field to override:
compilers:
- path: /usr/bin/cc
as: clang
- path: /usr/bin/c++
as: clang
Filtering functionality based on the source file location.
Directory rules are evaluated in order, with the last matching rule determining inclusion/exclusion. Empty directories list means include everything.
Filtering functionality based on duplicate detection. Here you can define which fields of the output file should be used in the duplicate detection. Two entries are duplicates when all of the configured fields match; the first occurrence is kept and later duplicates are dropped.
directory and file, so by default one entry is kept per source file per directory, regardless of the compiler arguments. A consequence is that when a single build compiles the same file from the same directory with different flags, only one entry is kept (the first seen); add arguments to match_on to keep a separate entry per configuration. Combined with --append (new entries first), the default means a rebuilt file's newest invocation replaces its stale entry.Output formatting configuration:
false for consumers that read only the command field.@file response-file references in each entry's arguments with the file's tokenized contents (resolved relative to the compiler's working directory, expanded recursively, MSVC/clang-cl using Windows quoting and other compilers using GCC/Clang quoting). Disabled by default, in which case an @file argument is recorded verbatim. Missing or unreadable files are left literal with a warning.CPATH, C_INCLUDE_PATH, CPLUS_INCLUDE_PATH, OBJC_INCLUDE_PATH) become include flags, and MSVC's CL / _CL_ become leading / trailing options. Enabled by default. Set to false to record only the flags that appeared on the command line. (This is unrelated to the CC="gcc -std=c11" convention handled during interception.)If no configuration file is specified, Bear uses built-in defaults optimized for most use cases.
RUST_LOG : Controls the logging level for Bear's internal operations. This environment variable is essential for troubleshooting and debugging Bear's behavior.
Supported log levels (in order of verbosity):
- `error` - Only show critical errors
- `warn` - Show warnings and errors
- `info` - Show informational messages, warnings, and errors
- `debug` - Show detailed debugging information
Examples:
```
RUST_LOG=debug bear -- make all
RUST_LOG=info bear intercept -- cmake --build .
```
The configuration file bear.yml is searched in the following locations, in order:
./bear.yml
: The current working directory.
$XDG_CONFIG_HOME/bear.yml, $XDG_CONFIG_HOME/Bear/bear.yml (Unix)
: When $XDG_CONFIG_HOME is set.
$HOME/.config/bear.yml, $HOME/.config/Bear/bear.yml (Unix)
: When $XDG_CONFIG_HOME is unset.
%LOCALAPPDATA%\bear.yml, %LOCALAPPDATA%\Bear\bear.yml (Windows)
: When %LOCALAPPDATA% is set.
%APPDATA%\bear.yml, %APPDATA%\Bear\bear.yml (Windows)
: When %APPDATA% is set.
The first file found is loaded; remaining locations are not consulted.
Bear returns the exit status of the executed build command when running in combined or intercept mode. When the build command succeeds, Bear returns 0. When the build command fails, Bear returns the same non-zero exit code.
In semantic mode, Bear returns 0 on success and a non-zero exit code if semantic analysis fails.
If Bear itself encounters an internal error or crashes, it returns a non-zero exit code regardless of the build command's status.
The potential problems you can face with are: the build with and without Bear behaves differently or the output is empty.
Before reporting any issues, always run Bear with debug logging enabled:
RUST_LOG=debug bear -- your-build-command
This will provide detailed information about Bear's internal operations. And the debug output is essential for diagnosing problems and must be included in any bug reports.
The most common cause for empty outputs is that the build command did not execute any commands. The reason for that could be, because incremental builds not running the compilers if everything is up-to-date. Remember, Bear does not understand the build file (eg.: makefile), but intercepts the executed commands.
The other common cause for empty output is that the build has a "configure" step, which captures the compiler to build the project. In case of Bear is using the wrapper mode, it needs to run the configure step with Bear too (and discard that output), before run the build with Bear.
When the build runs compilers from a cross-compilation SDK and you see an
error like version 'GLIBC_2.33' not found (required by .../libexec.so),
Bear's preload library was built against a newer glibc than the SDK toolchain
provides. The library must be ABI-compatible not only with the host system
but also with the libc the intercepted compiler process loads from the SDK
sysroot. Build (or obtain) a Bear linked against a glibc no newer than the
SDK's. See the project wiki Troubleshooting page (LD_PRELOAD errors) for
diagnostic commands.
In wrapper mode, Bear accepts compiler environment variables that carry a trailing flag or two, matching the GNU Make convention:
CC="gcc -std=c11" make
CXX="clang++ -stdlib=libc++" make
CC="/usr/local/bin/gcc -m32" make
Bear splits the value on whitespace, resolves the first token as the
compiler, and rewrites the variable so the build still sees the flags
(CC=<wrapper_path> -std=c11).
This convention is a Unix / GNU Make inheritance; it applies when
bear -- make runs under sh/bash (including MSYS2, Git Bash, WSL on
Windows), not under native Windows build systems (MSBuild, nmake,
cmd, PowerShell), which do not consume CC/CXX from the
environment.
For anything beyond simple whitespace-separated tokens (flags containing
spaces, shell quoting, metacharacters, command substitutions), use
CFLAGS, CXXFLAGS, or LDFLAGS instead of packing it into CC. Those
variables are the portable channel for compilation flags and every build
system expects them. Bear does not parse or rewrite them; they reach the
compiler untouched.
Bear records valac invocations, producing one entry per valac
invocation (valac compiles all of a target's .vala/.gs sources
together as one translation unit, so the entry's file is the first
source and every source is kept in the command). Two things are worth
knowing:
vala-language-server might require the command-string form. It
reads the command field and ignores the arguments array, so build
the database with the array format disabled:
format:
entries:
use_array_format: false
Mixed C and Vala projects. valac transpiles to C and then invokes a
C compiler on the generated C, so the database also contains entries for
that generated C. A C language server such as clangd indexes every entry
and will emit unknown-argument noise on the valac entries. Exclude the
Vala sources on the clangd side with a .clangd file:
If:
PathMatch: .*\.vala
Diagnostics:
Suppress: '*'
There could be many reasons for any of these failures. When seeking help:
RUST_LOG=debug) in your reportCopyright (C) 2012-2026 by László Nagy https://github.com/rizsotto/Bear