3 Evan Martin <martine@danga.com>
9 Ninja is yet another build system. It takes as input the
10 interdependencies of files (typically source code and output
11 executables) and orchestrates building them, _quickly_.
13 Ninja joins a sea of other build systems. Its distinguishing goal is
14 to be fast. It is born from
15 http://neugierig.org/software/chromium/notes/2011/02/ninja.html[my
16 work on the Chromium browser project], which has over 30,000 source
17 files and whose other build systems (including one built from custom
18 non-recursive Makefiles) would take ten seconds to start building
19 after changing one file. Ninja is under a second.
21 Philosophical overview
22 ~~~~~~~~~~~~~~~~~~~~~~
24 Where other build systems are high-level languages, Ninja aims to be
27 Build systems get slow when they need to make decisions. When you are
28 in a edit-compile cycle you want it to be as fast as possible -- you
29 want the build system to do the minimum work necessary to figure out
30 what needs to be built immediately.
32 Ninja contains the barest functionality necessary to describe
33 arbitrary dependency graphs. Its lack of syntax makes it impossible
34 to express complex decisions.
36 Instead, Ninja is intended to be used with a separate program
37 generating its input files. The generator program (like the
38 `./configure` found in autotools projects) can analyze system
39 dependencies and make as many decisions as possible up front so that
40 incremental builds stay fast. Going beyond autotools, even build-time
41 decisions like "which compiler flags should I use?" or "should I
42 build a debug or release-mode binary?" belong in the `.ninja` file
48 Here are the design goals of Ninja:
50 * very fast (i.e., instant) incremental builds, even for very large
53 * very little policy about how code is built. Different projects and
54 higher-level build systems have different opinions about how code
55 should be built; for example, should built objects live alongside
56 the sources or should all build output go into a separate directory?
57 Is there a "package" rule that builds a distributable package of
58 the project? Sidestep these decisions by trying to allow either to
59 be implemented, rather than choosing, even if that results in
62 * get dependencies correct, and in particular situations that are
63 difficult to get right with Makefiles (e.g. outputs need an implicit
64 dependency on the command line used to generate them; to build C
65 source code you need to use gcc's `-M` flags for header
68 * when convenience and speed are in conflict, prefer speed.
70 Some explicit _non-goals_:
72 * convenient syntax for writing build files by hand. _You should
73 generate your ninja files using another program_. This is how we
74 can sidestep many policy decisions.
76 * built-in rules. _Out of the box, Ninja has no rules for
77 e.g. compiling C code._
79 * build-time customization of the build. _Options belong in
80 the program that generates the ninja files_.
82 * build-time decision-making ability such as conditionals or search
83 paths. _Making decisions is slow._
85 To restate, Ninja is faster than other build systems because it is
86 painfully simple. You must tell Ninja exactly what to do when you
87 create your project's `.ninja` files.
92 Ninja is closest in spirit and functionality to Make, relying on
93 simple dependencies between file timestamps.
95 But fundamentally, make has a lot of _features_: suffix rules,
96 functions, built-in rules that e.g. search for RCS files when building
97 source. Make's language was designed to be written by humans. Many
98 projects find make alone adequate for their build problems.
100 In contrast, Ninja has almost no features; just those necessary to get
101 builds correct while punting most complexity to generation of the
102 ninja input files. Ninja by itself is unlikely to be useful for most
105 Here are some of the features Ninja adds to Make. (These sorts of
106 features can often be implemented using more complicated Makefiles,
107 but they are not part of make itself.)
109 * Ninja has special support for discovering extra dependencies at build
110 time, making it easy to get <<ref_headers,header dependencies>>
111 correct for C/C++ code.
113 * A build edge may have multiple outputs.
115 * Outputs implicitly depend on the command line that was used to generate
116 them, which means that changing e.g. compilation flags will cause
117 the outputs to rebuild.
119 * Output directories are always implicitly created before running the
120 command that relies on them.
122 * Rules can provide shorter descriptions of the command being run, so
123 you can print e.g. `CC foo.o` instead of a long command line while
126 * Builds are always run in parallel, based by default on the number of
127 CPUs your system has. Underspecified build dependencies will result
130 * Command output is always buffered. This means commands running in
131 parallel don't interleave their output, and when a command fails we
132 can print its failure output next to the full command line that
133 produced the failure.
136 Using Ninja for your project
137 ----------------------------
139 Ninja currently works on Unix-like systems and Windows. It's seen the
140 most testing on Linux (and has the best performance there) but it runs
141 fine on Mac OS X and FreeBSD.
143 If your project is small, Ninja's speed impact is likely unnoticeable.
144 (However, even for small projects it sometimes turns out that Ninja's
145 limited syntax forces simpler build rules that result in faster
146 builds.) Another way to say this is that if you're happy with the
147 edit-compile cycle time of your project already then Ninja won't help.
149 There are many other build systems that are more user-friendly or
150 featureful than Ninja itself. For some recommendations: the Ninja
151 author found http://gittup.org/tup/[the tup build system] influential
152 in Ninja's design, and thinks https://github.com/apenwarr/redo[redo]'s
153 design is quite clever.
155 Ninja's benefit comes from using it in conjunction with a smarter
158 http://code.google.com/p/gyp/[gyp]:: The meta-build system used to
159 generate build files for Google Chrome and related projects (v8,
160 node.js). gyp can generate Ninja files for all platforms supported by
162 http://code.google.com/p/chromium/wiki/NinjaBuild[Chromium Ninja
163 documentation for more details].
165 http://www.cmake.org/[CMake]:: A widely used meta-build system that
166 can generate Ninja files on Linux as of CMake version 2.8.8. (There
167 is some Mac and Windows support -- http://www.reactos.org[ReactOS]
168 uses Ninja on Windows for their buildbots, but those platforms are not
169 yet officially supported by CMake as the full test suite doesn't
172 others:: Ninja ought to fit perfectly into other meta-build software
173 like http://industriousone.com/premake[premake]. If you do this work,
179 Run `ninja`. By default, it looks for a file named `build.ninja` in
180 the current directory and builds all out-of-date targets. You can
181 specify which targets (files) to build as command line arguments.
183 There is also a special syntax `target^` for specifying a target
184 as the first output of some rule containing the source you put in
185 the command line, if one exists. For example, if you specify target as
186 `foo.c^` then `foo.o` will get built (assuming you have those targets
187 in your build files).
189 `ninja -h` prints help output. Many of Ninja's flags intentionally
190 match those of Make; e.g `ninja -C build -j 20` changes into the
191 `build` directory and runs 20 build commands in parallel. (Note that
192 Ninja defaults to running commands in parallel anyway, so typically
193 you don't need to pass `-j`.)
196 Environment variables
197 ~~~~~~~~~~~~~~~~~~~~~
199 Ninja supports one environment variable to control its behavior:
200 `NINJA_STATUS`, the progress status printed before the rule being run.
202 Several placeholders are available:
204 `%s`:: The number of started edges.
205 `%t`:: The total number of edges that must be run to complete the build.
206 `%p`:: The percentage of started edges.
207 `%r`:: The number of currently running edges.
208 `%u`:: The number of remaining edges to start.
209 `%f`:: The number of finished edges.
210 `%o`:: Overall rate of finished edges per second
211 `%c`:: Current rate of finished edges per second (average over builds
212 specified by `-j` or its default)
213 `%e`:: Elapsed time in seconds. _(Available since Ninja 1.2.)_
214 `%%`:: A plain `%` character.
216 The default progress status is `"[%s/%t] "` (note the trailing space
217 to separate from the build rule). Another example of possible progress status
218 could be `"[%u/%r/%f] "`.
223 The `-t` flag on the Ninja command line runs some tools that we have
224 found useful during Ninja's development. The current tools are:
227 `query`:: dump the inputs and outputs of a given target.
229 `browse`:: browse the dependency graph in a web browser. Clicking a
230 file focuses the view on that file, showing inputs and outputs. This
231 feature requires a Python installation.
233 `graph`:: output a file in the syntax used by `graphviz`, a automatic
234 graph layout tool. Use it like:
237 ninja -t graph mytarget | dot -Tpng -ograph.png
240 In the Ninja source tree, `ninja graph.png`
241 generates an image for Ninja itself. If no target is given generate a
242 graph for all root targets.
244 `targets`:: output a list of targets either by rule or by depth. If used
245 like +ninja -t targets rule _name_+ it prints the list of targets
246 using the given rule to be built. If no rule is given, it prints the source
247 files (the leaves of the graph). If used like
248 +ninja -t targets depth _digit_+ it
249 prints the list of targets in a depth-first manner starting by the root
250 targets (the ones with no outputs). Indentation is used to mark dependencies.
251 If the depth is zero it prints all targets. If no arguments are provided
252 +ninja -t targets depth 1+ is assumed. In this mode targets may be listed
253 several times. If used like this +ninja -t targets all+ it
254 prints all the targets available without indentation and it is faster
255 than the _depth_ mode.
257 `commands`:: given a list of targets, print a list of commands which, if
258 executed in order, may be used to rebuild those targets, assuming that all
259 output files are out of date.
261 `clean`:: remove built files. By default it removes all built files
262 except for those created by the generator. Adding the `-g` flag also
263 removes built files created by the generator (see <<ref_rule,the rule
264 reference for the +generator+ attribute>>). Additional arguments are
265 targets, which removes the given targets and recursively all files
268 If used like +ninja -t clean -r _rules_+ it removes all files built using
271 Files created but not referenced in the graph are not removed. This
272 tool takes in account the +-v+ and the +-n+ options (note that +-n+
275 `compdb`:: given a list of rules, each of which is expected to be a
276 C family language compiler rule whose first input is the name of the
277 source file, prints on standard output a compilation database in the
278 http://clang.llvm.org/docs/JSONCompilationDatabase.html[JSON format] expected
279 by the Clang tooling interface.
280 _Available since Ninja 1.2._
283 Writing your own Ninja files
284 ----------------------------
286 The remainder of this manual is only useful if you are constructing
287 Ninja files yourself: for example, if you're writing a meta-build
288 system or supporting a new language.
293 Ninja evaluates a graph of dependencies between files, and runs
294 whichever commands are necessary to make your build target up to date
295 as determined by file modification times. If you are familiar with
296 Make, Ninja is very similar.
298 A build file (default name: `build.ninja`) provides a list of _rules_
299 -- short names for longer commands, like how to run the compiler --
300 along with a list of _build_ statements saying how to build files
301 using the rules -- which rule to apply to which inputs to produce
304 Conceptually, `build` statements describe the dependency graph of your
305 project, while `rule` statements describe how to generate the files
306 along a given edge of the graph.
311 Here's a basic `.ninja` file that demonstrates most of the syntax.
312 It will be used as an example for the following sections.
314 ---------------------------------
318 command = gcc $cflags -c $in -o $out
320 build foo.o: cc foo.c
321 ---------------------------------
325 Despite the non-goal of being convenient to write by hand, to keep
326 build files readable (debuggable), Ninja supports declaring shorter
327 reusable names for strings. A declaration like the following
333 can be used on the right side of an equals sign, dereferencing it with
334 a dollar sign, like this:
338 command = gcc $cflags -c $in -o $out
341 Variables can also be referenced using curly braces like `${in}`.
343 Variables might better be called "bindings", in that a given variable
344 cannot be changed, only shadowed. There is more on how shadowing works
345 later in this document.
350 Rules declare a short name for a command line. They begin with a line
351 consisting of the `rule` keyword and a name for the rule. Then
352 follows an indented set of `variable = value` lines.
354 The basic example above declares a new rule named `cc`, along with the
355 command to run. In the context of a rule, the `command` variable
356 defines the command to run, `$in` expands to the list of
357 input files (`foo.c`), and `$out` to the output files (`foo.o`) for the
358 command. A full list of special variables is provided in
359 <<ref_rule,the reference>>.
364 Build statements declare a relationship between input and output
365 files. They begin with the `build` keyword, and have the format
366 +build _outputs_: _rulename_ _inputs_+. Such a declaration says that
367 all of the output files are derived from the input files. When the
368 output files are missing or when the inputs change, Ninja will run the
369 rule to regenerate the outputs.
371 The basic example above describes how to build `foo.o`, using the `cc`
374 In the scope of a `build` block (including in the evaluation of its
375 associated `rule`), the variable `$in` is the list of inputs and the
376 variable `$out` is the list of outputs.
378 A build statement may be followed by an indented set of `key = value`
379 pairs, much like a rule. These variables will shadow any variables
380 when evaluating the variables in the command. For example:
383 cflags = -Wall -Werror
385 command = gcc $cflags -c $in -o $out
387 # If left unspecified, builds get the outer $cflags.
388 build foo.o: cc foo.c
390 # But you can shadow variables like cflags for a particular build.
391 build special.o: cc special.c
394 # The variable was only shadowed for the scope of special.o;
395 # Subsequent build lines get the outer (original) cflags.
396 build bar.o: cc bar.c
400 For more discussion of how scoping works, consult <<ref_scope,the
403 If you need more complicated information passed from the build
404 statement to the rule (for example, if the rule needs "the file
405 extension of the first input"), pass that through as an extra
406 variable, like how `cflags` is passed above.
408 If the top-level Ninja file is specified as an output of any build
409 statement and it is out of date, Ninja will rebuild and reload it
410 before building the targets requested by the user.
412 Generating Ninja files from code
413 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
415 `misc/ninja_syntax.py` in the Ninja distribution is a tiny Python
416 module to facilitate generating Ninja files. It allows you to make
417 Python calls like `ninja.rule(name='foo', command='bar',
418 depfile='$out.d')` and it will generate the appropriate syntax. Feel
419 free to just inline it into your project's build system if it's
429 The special rule name `phony` can be used to create aliases for other
430 targets. For example:
433 build foo: phony some/file/in/a/faraway/subdir/foo
436 This makes `ninja foo` build the longer path. Semantically, the
437 `phony` rule is equivalent to a plain rule where the `command` does
438 nothing, but phony rules are handled specially in that they aren't
439 printed when run, logged (see below), nor do they contribute to the
440 command count printed as part of the build process.
442 `phony` can also be used to create dummy targets for files which
443 may not exist at build time. If a phony build statement is written
444 without any dependencies, the target will be considered out of date if
445 it does not exist. Without a phony build statement, Ninja will report
446 an error if the file does not exist and is required by the build.
449 Default target statements
450 ~~~~~~~~~~~~~~~~~~~~~~~~~
452 By default, if no targets are specified on the command line, Ninja
453 will build every output that is not named as an input elsewhere.
454 You can override this behavior using a default target statement.
455 A default target statement causes Ninja to build only a given subset
456 of output files if none are specified on the command line.
458 Default target statements begin with the `default` keyword, and have
459 the format +default _targets_+. A default target statement must appear
460 after the build statement that declares the target as an output file.
461 They are cumulative, so multiple statements may be used to extend
462 the list of default targets. For example:
469 This causes Ninja to build the `foo`, `bar` and `baz` targets by
477 For each built file, Ninja keeps a log of the command used to build
478 it. Using this log Ninja can know when an existing output was built
479 with a different command line than the build files specify (i.e., the
480 command line changed) and knows to rebuild the file.
482 The log file is kept in the build root in a file called `.ninja_log`.
483 If you provide a variable named `builddir` in the outermost scope,
484 `.ninja_log` will be kept in that directory instead.
488 Version compatibility
489 ~~~~~~~~~~~~~~~~~~~~~
491 _Available since Ninja 1.2._
493 Ninja version labels follow the standard major.minor.patch format,
494 where the major version is increased on backwards-incompatible
495 syntax/behavioral changes and the minor version is increased on new
496 behaviors. Your `build.ninja` may declare a variable named
497 `ninja_required_version` that asserts the minimum Ninja version
498 required to use the generated file. For example,
501 ninja_required_version = 1.1
504 declares that the build file relies on some feature that was
505 introduced in Ninja 1.1 (perhaps the `pool` syntax), and that
506 Ninja 1.1 or greater must be used to build. Unlike other Ninja
507 variables, this version requirement is checked immediately when
508 the variable is encountered in parsing, so it's best to put it
509 at the top of the build file.
511 Ninja always warns if the major versions of Ninja and the
512 `ninja_required_version` don't match; a major version change hasn't
513 come up yet so it's difficult to predict what behavior might be
517 C/C++ header dependencies
518 ~~~~~~~~~~~~~~~~~~~~~~~~~
520 To get C/C++ header dependencies (or any other build dependency that
521 works in a similar way) correct Ninja has some extra functionality.
523 The problem with headers is that the full list of files that a given
524 source file depends on can only be discovered by the compiler:
525 different preprocessor defines and include paths cause different files
526 to be used. Some compilers can emit this information while building,
527 and Ninja can use that to get its dependencies perfect.
529 Consider: if the file has never been compiled, it must be built anyway,
530 generating the header dependencies as a side effect. If any file is
531 later modified (even in a way that changes which headers it depends
532 on) the modification will cause a rebuild as well, keeping the
533 dependencies up to date.
535 When loading these special dependencies, Ninja implicitly adds extra
536 build edges such that it is not an error if the listed dependency is
537 missing. This allows you to delete a header file and rebuild without
538 the build aborting due to a missing input.
543 `gcc` (and other compilers like `clang`) support emitting dependency
544 information in the syntax of a Makefile. (Any command that can write
545 dependencies in this form can be used, not just `gcc`.)
547 To bring this information into Ninja requires cooperation. On the
548 Ninja side, the `depfile` attribute on the `build` must point to a
549 path where this data is written. (Ninja only supports the limited
550 subset of the Makefile syntax emitted by compilers.) Then the command
551 must know to write dependencies into the `depfile` path.
552 Use it like in the following example:
557 command = gcc -MMD -MF $out.d [other gcc flags here]
560 The `-MMD` flag to `gcc` tells it to output header dependencies, and
561 the `-MF` flag tells it where to write them.
566 _(Available since Ninja 1.3.)_
568 It turns out that for large projects (and particularly on Windows,
569 where the file system is slow) loading these dependency files on
572 Ninja 1.3 can instead process dependencies just after they're generated
573 and save a compacted form of the same information in a Ninja-internal
576 Ninja supports this processing in two forms.
578 1. `deps = gcc` specifies that the tool outputs `gcc`-style dependencies
579 in the form of Makefiles. Adding this to the above example will
580 cause Ninja to process the `depfile` immediately after the
581 compilation finishes, then delete the `.d` file (which is only used
584 2. `deps = msvc` specifies that the tool outputs header dependencies
585 in the form produced by Visual Studio's compiler's
586 http://msdn.microsoft.com/en-us/library/hdkef6tk(v=vs.90).aspx[`/showIncludes`
587 flag]. Briefly, this means the tool outputs specially-formatted lines
588 to its stdout. Ninja then filters these lines from the displayed
589 output. No `depfile` attribute is necessary, but the localized string
590 in front of the the header file path. For instance
591 `msvc_deps_prefix = Note: including file: `
592 for a English Visual Studio (the default). Should be globally defined.
595 msvc_deps_prefix = Note: including file:
598 command = cl /showIncludes -c $in /Fo$out
605 _Available since Ninja 1.1._
607 Pools allow you to allocate one or more rules or edges a finite number
608 of concurrent jobs which is more tightly restricted than the default
611 This can be useful, for example, to restrict a particular expensive rule
612 (like link steps for huge executables), or to restrict particular build
613 statements which you know perform poorly when run concurrently.
615 Each pool has a `depth` variable which is specified in the build file.
616 The pool is then referred to with the `pool` variable on either a rule
617 or a build statement.
619 No matter what pools you specify, ninja will never run more concurrent jobs
620 than the default parallelism, or the number of jobs specified on the command
624 # No more than 4 links at a time.
628 # No more than 1 heavy object at a time.
629 pool heavy_object_pool
639 # The link_pool is used here. Only 4 links will run concurrently.
640 build foo.exe: link input.obj
642 # A build statement can be exempted from its rule's pool by setting an
643 # empty pool. This effectively puts the build statement back into the default
644 # pool, which has infinite depth.
645 build other.exe: link input.obj
648 # A build statement can specify a pool directly.
649 # Only one of these builds will run at a time.
650 build heavy_object1.obj: cc heavy_obj1.cc
651 pool = heavy_object_pool
652 build heavy_object2.obj: cc heavy_obj2.cc
653 pool = heavy_object_pool
660 _Available since Ninja 1.5._
662 There exists a pre-defined pool named `console` with a depth of 1. It has
663 the special property that any task in the pool has direct access to the
664 standard input, output and error streams provided to Ninja, which are
665 normally connected to the user's console (hence the name) but could be
666 redirected. This can be useful for interactive tasks or long-running tasks
667 which produce status updates on the console (such as test suites).
669 While a task in the `console` pool is running, Ninja's regular output (such
670 as progress status and output from concurrent tasks) is buffered until
676 A file is a series of declarations. A declaration can be one of:
678 1. A rule declaration, which begins with +rule _rulename_+, and
679 then has a series of indented lines defining variables.
681 2. A build edge, which looks like +build _output1_ _output2_:
682 _rulename_ _input1_ _input2_+. +
683 Implicit dependencies may be tacked on the end with +|
684 _dependency1_ _dependency2_+. +
685 Order-only dependencies may be tacked on the end with +||
686 _dependency1_ _dependency2_+. (See <<ref_dependencies,the reference on
689 3. Variable declarations, which look like +_variable_ = _value_+.
691 4. Default target statements, which look like +default _target1_ _target2_+.
693 5. References to more files, which look like +subninja _path_+ or
694 +include _path_+. The difference between these is explained below
695 <<ref_scope,in the discussion about scoping>>.
697 6. A pool declaration, which looks like +pool _poolname_+. Pools are explained
698 <<ref_pool, in the section on pools>>.
703 Ninja is mostly encoding agnostic, as long as the bytes Ninja cares
704 about (like slashes in paths) are ASCII. This means e.g. UTF-8 or
705 ISO-8859-1 input files ought to work.
707 Comments begin with `#` and extend to the end of the line.
709 Newlines are significant. Statements like `build foo bar` are a set
710 of space-separated tokens that end at the newline. Newlines and
711 spaces within a token must be escaped.
713 There is only one escape character, `$`, and it has the following
717 `$` followed by a newline:: escape the newline (continue the current line
718 across a line break).
720 `$` followed by text:: a variable reference.
722 `${varname}`:: alternate syntax for `$varname`.
724 `$` followed by space:: a space. (This is only necessary in lists of
725 paths, where a space would otherwise separate filenames. See below.)
727 `$:` :: a colon. (This is only necessary in `build` lines, where a colon
728 would otherwise terminate the list of outputs.)
730 `$$`:: a literal `$`.
732 A `build` or `default` statement is first parsed as a space-separated
733 list of filenames and then each name is expanded. This means that
734 spaces within a variable will result in spaces in the expanded
739 build $spaced/baz other$ file: ...
740 # The above build line has two outputs: "foo bar/baz" and "other file".
743 In a `name = value` statement, whitespace at the beginning of a value
744 is always stripped. Whitespace at the beginning of a line after a
745 line continuation is also stripped.
748 two_words_with_one_space = foo $
750 one_word_with_no_space = foo$
754 Other whitespace is only significant if it's at the beginning of a
755 line. If a line is indented more than the previous one, it's
756 considered part of its parent's scope; if it is indented less than the
757 previous one, it closes the previous scope.
763 Two variables are significant when declared in the outermost file scope.
765 `builddir`:: a directory for some Ninja output files. See <<ref_log,the
766 discussion of the build log>>. (You can also store other build output
769 `ninja_required_version`:: the minimum version of Ninja required to process
770 the build correctly. See <<ref_versioning,the discussion of versioning>>.
777 A `rule` block contains a list of `key = value` declarations that
778 affect the processing of the rule. Here is a full list of special
781 `command` (_required_):: the command line to run. Each `rule` may
782 have only one `command` declaration. See <<ref_rule_command,the next
783 section>> for more details on quoting and executing multiple commands.
785 `depfile`:: path to an optional `Makefile` that contains extra
786 _implicit dependencies_ (see <<ref_dependencies,the reference on
787 dependency types>>). This is explicitly to support C/C++ header
788 dependencies; see <<ref_headers,the full discussion>>.
790 `deps`:: _(Available since Ninja 1.3.)_ if present, must be one of
791 `gcc` or `msvc` to specify special dependency processing. See
792 <<ref_headers,the full discussion>>. The generated database is
793 stored as `.ninja_deps` in the `builddir`, see <<ref_toplevel,the
794 discussion of `builddir`>>.
796 `msvc_deps_prefix`:: _(Available since Ninja 1.5.)_ defines the string
797 which should be stripped from msvc's /showIncludes output. Only
798 needed when `deps = msvc` and no English Visual Studio version is used.
800 `description`:: a short description of the command, used to pretty-print
801 the command as it's running. The `-v` flag controls whether to print
802 the full command or its description; if a command fails, the full command
803 line will always be printed before the command's output.
805 `generator`:: if present, specifies that this rule is used to
806 re-invoke the generator program. Files built using `generator`
807 rules are treated specially in two ways: firstly, they will not be
808 rebuilt if the command line changes; and secondly, they are not
811 `in`:: the space-separated list of files provided as inputs to the build line
812 referencing this `rule`, shell-quoted if it appears in commands. (`$in` is
813 provided solely for convenience; if you need some subset or variant of this
814 list of files, just construct a new variable with that list and use
817 `in_newline`:: the same as `$in` except that multiple inputs are
818 separated by newlines rather than spaces. (For use with
819 `$rspfile_content`; this works around a bug in the MSVC linker where
820 it uses a fixed-size buffer for processing input.)
822 `out`:: the space-separated list of files provided as outputs to the build line
823 referencing this `rule`, shell-quoted if it appears in commands.
825 `restat`:: if present, causes Ninja to re-stat the command's outputs
826 after execution of the command. Each output whose modification time
827 the command did not change will be treated as though it had never
828 needed to be built. This may cause the output's reverse
829 dependencies to be removed from the list of pending build actions.
831 `rspfile`, `rspfile_content`:: if present (both), Ninja will use a
832 response file for the given command, i.e. write the selected string
833 (`rspfile_content`) to the given file (`rspfile`) before calling the
834 command and delete the file after successful execution of the
837 This is particularly useful on Windows OS, where the maximal length of
838 a command line is limited and response files must be used instead.
840 Use it like in the following example:
844 command = link.exe /OUT$out [usual link flags here] @$out.rsp
846 rspfile_content = $in
848 build myapp.exe: link a.obj b.obj [possibly many other .obj files]
852 Interpretation of the `command` variable
853 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
854 Fundamentally, command lines behave differently on Unixes and Windows.
856 On Unixes, commands are arrays of arguments. The Ninja `command`
857 variable is passed directly to `sh -c`, which is then responsible for
858 interpreting that string into an argv array. Therefore the quoting
859 rules are those of the shell, and you can use all the normal shell
860 operators, like `&&` to chain multiple commands, or `VAR=value cmd` to
861 set environment variables.
863 On Windows, commands are strings, so Ninja passes the `command` string
864 directly to `CreateProcess`. (In the common case of simply executing
865 a compiler this means there is less overhead.) Consequently the
866 quoting rules are deterimined by the called program, which on Windows
867 are usually provided by the C library. If you need shell
868 interpretation of the command (such as the use of `&&` to chain
869 multiple commands), make the command execute the Windows shell by
870 prefixing the command with `cmd /c`.
876 There are three types of build dependencies which are subtly different.
878 1. _Explicit dependencies_, as listed in a build line. These are
879 available as the `$in` variable in the rule. Changes in these files
880 cause the output to be rebuilt; if these file are missing and
881 Ninja doesn't know how to build them, the build is aborted.
883 This is the standard form of dependency to be used for e.g. the
884 source file of a compile command.
886 2. _Implicit dependencies_, either as picked up from
887 a `depfile` attribute on a rule or from the syntax +| _dep1_
888 _dep2_+ on the end of a build line. The semantics are identical to
889 explicit dependencies, the only difference is that implicit dependencies
890 don't show up in the `$in` variable.
892 This is for expressing dependencies that don't show up on the
893 command line of the command; for example, for a rule that runs a
894 script, the script itself should be an implicit dependency, as
895 changes to the script should cause the output to rebuild.
897 Note that dependencies as loaded through depfiles have slightly different
898 semantics, as described in the <<ref_rule,rule reference>>.
900 3. _Order-only dependencies_, expressed with the syntax +|| _dep1_
901 _dep2_+ on the end of a build line. When these are out of date, the
902 output is not rebuilt until they are built, but changes in order-only
903 dependencies alone do not cause the output to be rebuilt.
905 Order-only dependencies can be useful for bootstrapping dependencies
906 that are only discovered during build time: for example, to generate a
907 header file before starting a subsequent compilation step. (Once the
908 header is used in compilation, a generated dependency file will then
909 express the implicit dependency.)
914 Variables are expanded in paths (in a `build` or `default` statement)
915 and on the right side of a `name = value` statement.
917 When a `name = value` statement is evaluated, its right-hand side is
918 expanded immediately (according to the below scoping rules), and
919 from then on `$name` expands to the static string as the result of the
920 expansion. It is never the case that you'll need to "double-escape" a
921 value to prevent it from getting expanded twice.
923 All variables are expanded immediately as they're encountered in parsing,
924 with one important exception: variables in `rule` blocks are expanded
925 when the rule is _used_, not when it is declared. In the following
926 example, the `demo` rule prints "this is a demo of bar".
930 command = echo "this is a demo of $foo"
937 Evaluation and scoping
938 ~~~~~~~~~~~~~~~~~~~~~~
940 Top-level variable declarations are scoped to the file they occur in.
942 Rule declarations are also scoped to the file they occur in.
943 _(Available since Ninja 1.6)_
945 The `subninja` keyword, used to include another `.ninja` file,
946 introduces a new scope. The included `subninja` file may use the
947 variables and rules from the parent file, and shadow their values for the file's
948 scope, but it won't affect values of the variables in the parent.
950 To include another `.ninja` file in the current scope, much like a C
951 `#include` statement, use `include` instead of `subninja`.
953 Variable declarations indented in a `build` block are scoped to the
954 `build` block. The full lookup order for a variable expanded in a
955 `build` block (or the `rule` is uses) is:
957 1. Special built-in variables (`$in`, `$out`).
959 2. Build-level variables from the `build` block.
961 3. Rule-level variables from the `rule` block (i.e. `$command`).
962 (Note from the above discussion on expansion that these are
963 expanded "late", and may make use of in-scope bindings like `$in`.)
965 4. File-level variables from the file that the `build` line was in.
967 5. Variables from the file that included that file using the