Technical documentation of the run-command API.
Signed-off-by: Johannes Sixt <johannes.sixt@telecom.at> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Johannes Sixt committed
Feb 18, 2008 at 20:23 UTC
b9dfe51c96cfc258cb0bc41433853892c35e0beb
1 file changed
+166
-5
Documentation/technical/api-run-command.txt
+166
-5
@@ -1,10 +1,171 @@
1
run-command API
2
===============
3
4
-Talk about <run-command.h>, and things like:
4
+The run-command API offers a versatile tool to run sub-processes with
5
+redirected input and output as well as with a modified environment
6
+and an alternate current directory.
7
6
-* Environment the command runs with (e.g. GIT_DIR);
7
-* File descriptors and pipes;
8
-* Exit status;
8
+A similar API offers the capability to run a function asynchronously,
9
+which is primarily used to capture the output that the function
10
+produces in the caller in order to process it.
11
10
-(Hannes, Dscho, Shawn)
12
+
13
+Functions
14
+---------
15
+
16
+`start_command`::
17
+
18
+ Start a sub-process. Takes a pointer to a `struct child_process`
19
+ that specifies the details and returns pipe FDs (if requested).
20
+ See below for details.
21
+
22
+`finish_command`::
23
+
24
+ Wait for the completion of a sub-process that was started with
25
+ start_command().
26
+
27
+`run_command`::
28
+
29
+ A convenience function that encapsulates a sequence of
30
+ start_command() followed by finish_command(). Takes a pointer
31
+ to a `struct child_process` that specifies the details.
32
+
33
+`run_command_v_opt`, `run_command_v_opt_dir`, `run_command_v_opt_cd_env`::
34
+
35
+ Convenience functions that encapsulate a sequence of
36
+ start_command() followed by finish_command(). The argument argv
37
+ specifies the program and its arguments. The argument opt is zero
38
+ or more of the flags `RUN_COMMAND_NO_STDIN`, `RUN_GIT_CMD`, or
39
+ `RUN_COMMAND_STDOUT_TO_STDERR` that correspond to the members
40
+ .no_stdin, .git_cmd, .stdout_to_stderr of `struct child_process`.
41
+ The argument dir corresponds the member .dir. The argument env
42
+ corresponds to the member .env.
43
+
44
+`start_async`::
45
+
46
+ Run a function asynchronously. Takes a pointer to a `struct
47
+ async` that specifies the details and returns a pipe FD
48
+ from which the caller reads. See below for details.
49
+
50
+`finish_async`::
51
+
52
+ Wait for the completeion of an asynchronous function that was
53
+ started with start_async().
54
+
55
+
56
+Data structures
57
+---------------
58
+
59
+* `struct child_process`
60
+
61
+This describes the arguments, redirections, and environment of a
62
+command to run in a sub-process.
63
+
64
+The caller:
65
+
66
+1. allocates and clears (memset(&chld, '0', sizeof(chld));) a
67
+ struct child_process variable;
68
+2. initializes the members;
69
+3. calls start_command();
70
+4. processes the data;
71
+5. closes file descriptors (if necessary; see below);
72
+6. calls finish_command().
73
+
74
+The .argv member is set up as an array of string pointers (NULL
75
+terminated), of which .argv[0] is the program name to run (usually
76
+without a path). If the command to run is a git command, set argv[0] to
77
+the command name without the 'git-' prefix and set .git_cmd = 1.
78
+
79
+The members .in, .out, .err are used to redirect stdin, stdout,
80
+stderr as follows:
81
+
82
+. Specify 0 to request no special redirection. No new file descriptor
83
+ is allocated. The child process simply inherits the channel from the
84
+ parent.
85
+
86
+. Specify -1 to have a pipe allocated; start_command() replaces -1
87
+ by the pipe FD in the following way:
88
+
89
+ .in: Returns the writable pipe end into which the caller writes;
90
+ the readable end of the pipe becomes the child's stdin.
91
+
92
+ .out, .err: Returns the readable pipe end from which the caller
93
+ reads; the writable end of the pipe end becomes child's
94
+ stdout/stderr.
95
+
96
+ The caller of start_command() must close the so returned FDs
97
+ after it has completed reading from/writing to it!
98
+
99
+. Specify a file descriptor > 0 to be used by the child:
100
+
101
+ .in: The FD must be readable; it becomes child's stdin.
102
+ .out: The FD must be writable; it becomes child's stdout.
103
+ .err > 0 is not supported.
104
+
105
+ The specified FD is closed by start_command(), even if it fails to
106
+ run the sub-process!
107
+
108
+. Special forms of redirection are available by setting these members
109
+ to 1:
110
+
111
+ .no_stdin, .no_stdout, .no_stderr: The respective channel is
112
+ redirected to /dev/null.
113
+
114
+ .stdout_to_stderr: stdout of the child is redirected to the
115
+ parent's stderr (i.e. *not* to what .err or
116
+ .no_stderr specify).
117
+
118
+To modify the environment of the sub-process, specify an array of
119
+string pointers (NULL terminated) in .env:
120
+
121
+. If the string is of the form "VAR=value", i.e. it contains '='
122
+ the variable is added to the child process's environment.
123
+
124
+. If the string does not contain '=', it names an environement
125
+ variable that will be removed from the child process's envionment.
126
+
127
+To specify a new initial working directory for the sub-process,
128
+specify it in the .dir member.
129
+
130
+
131
+* `struct async`
132
+
133
+This describes a function to run asynchronously, whose purpose is
134
+to produce output that the caller reads.
135
+
136
+The caller:
137
+
138
+1. allocates and clears (memset(&asy, '0', sizeof(asy));) a
139
+ struct async variable;
140
+2. initializes .proc and .data;
141
+3. calls start_async();
142
+4. processes the data by reading from the fd in .out;
143
+5. closes .out;
144
+6. calls finish_async().
145
+
146
+The function pointer in .proc has the following signature:
147
+
148
+ int proc(int fd, void *data);
149
+
150
+. fd specifies a writable file descriptor to which the function must
151
+ write the data that it produces. The function *must* close this
152
+ descriptor before it returns.
153
+
154
+. data is the value that the caller has specified in the .data member
155
+ of struct async.
156
+
157
+. The return value of the function is 0 on success and non-zero
158
+ on failure. If the function indicates failure, finish_async() will
159
+ report failure as well.
160
+
161
+
162
+There are serious restrictions on what the asynchronous function can do
163
+because this facility is implemented by a pipe to a forked process on
164
+UNIX, but by a thread in the same address space on Windows:
165
+
166
+. It cannot change the program's state (global variables, environment,
167
+ etc.) in a way that the caller notices; in other words, .out is the
168
+ only communication channel to the caller.
169
+
170
+. It must not change the program's state that the caller of the
171
+ facility also uses.