blob: 052867fc50d4aa4d5fd2f84f1c5f36ef37c939f0 [file] [log] [blame]
Fred Drake295da241998-08-10 19:42:37 +00001\section{\module{thread} ---
Fred Drake02e8c0f1999-04-21 18:01:14 +00002 Multiple threads of control}
Fred Drakeb91e9341998-07-23 17:59:49 +00003
Fred Drake02e8c0f1999-04-21 18:01:14 +00004\declaremodule{builtin}{thread}
Fred Drake295da241998-08-10 19:42:37 +00005\modulesynopsis{Create multiple threads of control within one interpreter.}
Fred Drakeb91e9341998-07-23 17:59:49 +00006
Guido van Rossum5fdeeea1994-01-02 01:22:07 +00007
8This module provides low-level primitives for working with multiple
Guido van Rossum6bb1adc1995-03-13 10:03:32 +00009threads (a.k.a.\ \dfn{light-weight processes} or \dfn{tasks}) --- multiple
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000010threads of control sharing their global data space. For
Guido van Rossum6bb1adc1995-03-13 10:03:32 +000011synchronization, simple locks (a.k.a.\ \dfn{mutexes} or \dfn{binary
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000012semaphores}) are provided.
Fred Drake61b04521998-01-20 05:52:23 +000013\index{light-weight processes}
14\index{processes, light-weight}
15\index{binary semaphores}
16\index{semaphores, binary}
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000017
Fred Drake7d428ec2002-10-22 15:06:49 +000018The module is optional. It is supported on Windows, Linux, SGI
Fred Drake65b32f71998-02-09 20:27:12 +000019IRIX, Solaris 2.x, as well as on systems that have a \POSIX{} thread
Guido van Rossum29692332002-12-30 22:34:10 +000020(a.k.a. ``pthread'') implementation. For systems lacking the \module{thread}
Fred Drake740f8002002-12-30 23:00:36 +000021module, the \refmodule[dummythread]{dummy_thread} module is available.
22It duplicates this module's interface and can be
23used as a drop-in replacement.
Fred Drake61b04521998-01-20 05:52:23 +000024\index{pthreads}
Fred Drakec37b65e2001-11-28 07:26:15 +000025\indexii{threads}{\POSIX}
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000026
27It defines the following constant and functions:
28
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000029\begin{excdesc}{error}
30Raised on thread-specific errors.
31\end{excdesc}
32
Guido van Rossum73d8bff1998-06-27 18:25:44 +000033\begin{datadesc}{LockType}
34This is the type of lock objects.
35\end{datadesc}
36
Fred Drake02e8c0f1999-04-21 18:01:14 +000037\begin{funcdesc}{start_new_thread}{function, args\optional{, kwargs}}
Guido van Rossum3c288632001-10-16 21:13:49 +000038Start a new thread and return its identifier. The thread executes the function
39\var{function} with the argument list \var{args} (which must be a tuple). The
40optional \var{kwargs} argument specifies a dictionary of keyword arguments.
41When the function returns, the thread silently exits. When the function
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000042terminates with an unhandled exception, a stack trace is printed and
43then the thread exits (but other threads continue to run).
44\end{funcdesc}
45
Kurt B. Kaiser0b1782f2003-07-08 17:07:20 +000046\begin{funcdesc}{interrupt_main}{}
47Raise a KeyboardInterrupt in the main thread. A subthread can use this
Neal Norwitza4d2b862003-07-09 12:41:55 +000048function to interrupt the main thread.
49\versionadded{2.3}
Kurt B. Kaiser0b1782f2003-07-08 17:07:20 +000050\end{funcdesc}
51
Guido van Rossum470be141995-03-17 16:07:09 +000052\begin{funcdesc}{exit}{}
Guido van Rossum73d8bff1998-06-27 18:25:44 +000053Raise the \exception{SystemExit} exception. When not caught, this
54will cause the thread to exit silently.
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000055\end{funcdesc}
56
Guido van Rossum470be141995-03-17 16:07:09 +000057%\begin{funcdesc}{exit_prog}{status}
58%Exit all threads and report the value of the integer argument
59%\var{status} as the exit status of the entire program.
Fred Drake02e8c0f1999-04-21 18:01:14 +000060%\strong{Caveat:} code in pending \keyword{finally} clauses, in this thread
Guido van Rossum470be141995-03-17 16:07:09 +000061%or in other threads, is not executed.
62%\end{funcdesc}
63
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000064\begin{funcdesc}{allocate_lock}{}
65Return a new lock object. Methods of locks are described below. The
66lock is initially unlocked.
67\end{funcdesc}
68
Guido van Rossumb8b264b1994-08-12 13:13:50 +000069\begin{funcdesc}{get_ident}{}
70Return the `thread identifier' of the current thread. This is a
71nonzero integer. Its value has no direct meaning; it is intended as a
72magic cookie to be used e.g. to index a dictionary of thread-specific
73data. Thread identifiers may be recycled when a thread exits and
74another thread is created.
75\end{funcdesc}
76
Fred Draked678cb71998-04-03 06:35:54 +000077
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000078Lock objects have the following methods:
79
Fred Draked678cb71998-04-03 06:35:54 +000080\begin{methoddesc}[lock]{acquire}{\optional{waitflag}}
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000081Without the optional argument, this method acquires the lock
82unconditionally, if necessary waiting until it is released by another
83thread (only one thread at a time can acquire a lock --- that's their
84reason for existence), and returns \code{None}. If the integer
Fred Draked61975e1998-11-30 16:26:50 +000085\var{waitflag} argument is present, the action depends on its
86value: if it is zero, the lock is only acquired if it can be acquired
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000087immediately without waiting, while if it is nonzero, the lock is
88acquired unconditionally as before. If an argument is present, the
Neal Norwitz6b353702002-04-09 18:15:00 +000089return value is \code{True} if the lock is acquired successfully,
90\code{False} if not.
Fred Draked678cb71998-04-03 06:35:54 +000091\end{methoddesc}
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000092
Fred Draked678cb71998-04-03 06:35:54 +000093\begin{methoddesc}[lock]{release}{}
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000094Releases the lock. The lock must have been acquired earlier, but not
95necessarily by the same thread.
Fred Draked678cb71998-04-03 06:35:54 +000096\end{methoddesc}
Guido van Rossum5fdeeea1994-01-02 01:22:07 +000097
Fred Draked678cb71998-04-03 06:35:54 +000098\begin{methoddesc}[lock]{locked}{}
Neal Norwitz6b353702002-04-09 18:15:00 +000099Return the status of the lock:\ \code{True} if it has been acquired by
100some thread, \code{False} if not.
Fred Draked678cb71998-04-03 06:35:54 +0000101\end{methoddesc}
Guido van Rossum5fdeeea1994-01-02 01:22:07 +0000102
Fred Drakeaf8a0151998-01-14 14:51:31 +0000103\strong{Caveats:}
Guido van Rossum5fdeeea1994-01-02 01:22:07 +0000104
105\begin{itemize}
106\item
107Threads interact strangely with interrupts: the
Fred Draked678cb71998-04-03 06:35:54 +0000108\exception{KeyboardInterrupt} exception will be received by an
Fred Drake02e8c0f1999-04-21 18:01:14 +0000109arbitrary thread. (When the \refmodule{signal}\refbimodindex{signal}
Fred Draked678cb71998-04-03 06:35:54 +0000110module is available, interrupts always go to the main thread.)
Guido van Rossum5fdeeea1994-01-02 01:22:07 +0000111
112\item
Fred Draked678cb71998-04-03 06:35:54 +0000113Calling \function{sys.exit()} or raising the \exception{SystemExit}
Fred Drake43b89b62000-04-05 15:00:38 +0000114exception is equivalent to calling \function{exit()}.
Guido van Rossum5fdeeea1994-01-02 01:22:07 +0000115
116\item
117Not all built-in functions that may block waiting for I/O allow other
Fred Draked678cb71998-04-03 06:35:54 +0000118threads to run. (The most popular ones (\function{time.sleep()},
119\method{\var{file}.read()}, \function{select.select()}) work as
120expected.)
Guido van Rossum5fdeeea1994-01-02 01:22:07 +0000121
Guido van Rossum3572d371997-08-14 19:51:26 +0000122\item
Fred Draked678cb71998-04-03 06:35:54 +0000123It is not possible to interrupt the \method{acquire()} method on a lock
124--- the \exception{KeyboardInterrupt} exception will happen after the
125lock has been acquired.
Guido van Rossum3572d371997-08-14 19:51:26 +0000126
127\item
128When the main thread exits, it is system defined whether the other
129threads survive. On SGI IRIX using the native thread implementation,
130they survive. On most other systems, they are killed without
Fred Draked678cb71998-04-03 06:35:54 +0000131executing \keyword{try} ... \keyword{finally} clauses or executing
132object destructors.
Fred Drake61b04521998-01-20 05:52:23 +0000133\indexii{threads}{IRIX}
Guido van Rossum3572d371997-08-14 19:51:26 +0000134
135\item
Fred Draked678cb71998-04-03 06:35:54 +0000136When the main thread exits, it does not do any of its usual cleanup
137(except that \keyword{try} ... \keyword{finally} clauses are honored),
138and the standard I/O files are not flushed.
Guido van Rossum3572d371997-08-14 19:51:26 +0000139
Guido van Rossum5fdeeea1994-01-02 01:22:07 +0000140\end{itemize}