18.3.select — Waiting for I/O completion

This module provides access to theselect() andpoll() functionsavailable in most operating systems,devpoll() available onSolaris and derivatives,epoll() available on Linux 2.5+ andkqueue() available on most BSD.Note that on Windows, it only works for sockets; on other operating systems,it also works for other file types (in particular, on Unix, it works on pipes).It cannot be used on regular files to determine whether a file has grown sinceit was last read.

Note

Theselectors module allows high-level and efficient I/Omultiplexing, built upon theselect module primitives. Users areencouraged to use theselectors module instead, unless they wantprecise control over the OS-level primitives used.

The module defines the following:

exceptionselect.error

A deprecated alias ofOSError.

Changed in version 3.3:FollowingPEP 3151, this class was made an alias ofOSError.

select.devpoll()

(Only supported on Solaris and derivatives.) Returns a/dev/pollpolling object; see section/dev/poll Polling Objects below for themethods supported by devpoll objects.

devpoll() objects are linked to the number of filedescriptors allowed at the time of instantiation. If your programreduces this value,devpoll() will fail. If your programincreases this value,devpoll() may return anincomplete list of active file descriptors.

The new file descriptor isnon-inheritable.

New in version 3.3.

Changed in version 3.4:The new file descriptor is now non-inheritable.

select.epoll(sizehint=-1,flags=0)

(Only supported on Linux 2.5.44 and newer.) Return an edge polling object,which can be used as Edge or Level Triggered interface for I/Oevents.sizehint andflags are deprecated and completely ignored.

See theEdge and Level Trigger Polling (epoll) Objects section below for the methods supported byepolling objects.

epoll objects support the context management protocol: when used in awith statement, the new file descriptor is automatically closedat the end of the block.

The new file descriptor isnon-inheritable.

Changed in version 3.3:Added theflags parameter.

Changed in version 3.4:Support for thewith statement was added.The new file descriptor is now non-inheritable.

Deprecated since version 3.4:Theflags parameter.select.EPOLL_CLOEXEC is used by default now.Useos.set_inheritable() to make the file descriptor inheritable.

select.poll()

(Not supported by all operating systems.) Returns a polling object, whichsupports registering and unregistering file descriptors, and then polling themfor I/O events; see sectionPolling Objects below for the methods supportedby polling objects.

select.kqueue()

(Only supported on BSD.) Returns a kernel queue object; see sectionKqueue Objects below for the methods supported by kqueue objects.

The new file descriptor isnon-inheritable.

Changed in version 3.4:The new file descriptor is now non-inheritable.

select.kevent(ident,filter=KQ_FILTER_READ,flags=KQ_EV_ADD,fflags=0,data=0,udata=0)

(Only supported on BSD.) Returns a kernel event object; see sectionKevent Objects below for the methods supported by kevent objects.

select.select(rlist,wlist,xlist[,timeout])

This is a straightforward interface to the Unixselect() system call.The first three arguments are sequences of ‘waitable objects’: eitherintegers representing file descriptors or objects with a parameterless methodnamedfileno() returning such an integer:

  • rlist: wait until ready for reading
  • wlist: wait until ready for writing
  • xlist: wait for an “exceptional condition” (see the manual page for whatyour system considers such a condition)

Empty sequences are allowed, but acceptance of three empty sequences isplatform-dependent. (It is known to work on Unix but not on Windows.) Theoptionaltimeout argument specifies a time-out as a floating point numberin seconds. When thetimeout argument is omitted the function blocks untilat least one file descriptor is ready. A time-out value of zero specifies apoll and never blocks.

The return value is a triple of lists of objects that are ready: subsets of thefirst three arguments. When the time-out is reached without a file descriptorbecoming ready, three empty lists are returned.

Among the acceptable object types in the sequences are Pythonfileobjects (e.g.sys.stdin, or objects returned byopen() oros.popen()), socket objects returned bysocket.socket(). You may also define awrapper class yourself,as long as it has an appropriatefileno() method (thatreally returns a file descriptor, not just a random integer).

Note

File objects on Windows are not acceptable, but sockets are. On Windows,the underlyingselect() function is provided by the WinSocklibrary, and does not handle file descriptors that don’t originate fromWinSock.

Changed in version 3.5:The function is now retried with a recomputed timeout when interrupted bya signal, except if the signal handler raises an exception (seePEP 475 for the rationale), instead of raisingInterruptedError.

select.PIPE_BUF

The minimum number of bytes which can be written without blocking to a pipewhen the pipe has been reported as ready for writing byselect(),poll() or another interface in this module. This doesn’t applyto other kind of file-like objects such as sockets.

This value is guaranteed by POSIX to be at least 512. Availability: Unix.

New in version 3.2.

18.3.1./dev/poll Polling Objects

Solaris and derivatives have/dev/poll. Whileselect() isO(highest file descriptor) andpoll() is O(number of filedescriptors),/dev/poll is O(active file descriptors).

/dev/poll behaviour is very close to the standardpoll()object.

devpoll.close()

Close the file descriptor of the polling object.

New in version 3.4.

devpoll.closed

True if the polling object is closed.

New in version 3.4.

devpoll.fileno()

Return the file descriptor number of the polling object.

New in version 3.4.

devpoll.register(fd[,eventmask])

Register a file descriptor with the polling object. Future calls to thepoll() method will then check whether the file descriptor has anypending I/O events.fd can be either an integer, or an object with afileno() method that returns an integer. File objectsimplementfileno(), so they can also be used as the argument.

eventmask is an optional bitmask describing the type of events you want tocheck for. The constants are the same that withpoll()object. The default value is a combination of the constantsPOLLIN,POLLPRI, andPOLLOUT.

Warning

Registering a file descriptor that’s already registered is not anerror, but the result is undefined. The appropriate action is tounregister or modify it first. This is an important differencecompared withpoll().

devpoll.modify(fd[,eventmask])

This method does anunregister() followed by aregister(). It is (a bit) more efficient that doing the sameexplicitly.

devpoll.unregister(fd)

Remove a file descriptor being tracked by a polling object. Just like theregister() method,fd can be an integer or an object with afileno() method that returns an integer.

Attempting to remove a file descriptor that was never registered issafely ignored.

devpoll.poll([timeout])

Polls the set of registered file descriptors, and returns a possibly-empty listcontaining(fd,event) 2-tuples for the descriptors that have events orerrors to report.fd is the file descriptor, andevent is a bitmask withbits set for the reported events for that descriptor —POLLIN forwaiting input,POLLOUT to indicate that the descriptor can be writtento, and so forth. An empty list indicates that the call timed out and no filedescriptors had any events to report. Iftimeout is given, it specifies thelength of time in milliseconds which the system will wait for events beforereturning. Iftimeout is omitted, -1, orNone, the call willblock until there is an event for this poll object.

Changed in version 3.5:The function is now retried with a recomputed timeout when interrupted bya signal, except if the signal handler raises an exception (seePEP 475 for the rationale), instead of raisingInterruptedError.

18.3.2. Edge and Level Trigger Polling (epoll) Objects

http://linux.die.net/man/4/epoll

eventmask

ConstantMeaning
EPOLLINAvailable for read
EPOLLOUTAvailable for write
EPOLLPRIUrgent data for read
EPOLLERRError condition happened on the assoc. fd
EPOLLHUPHang up happened on the assoc. fd
EPOLLETSet Edge Trigger behavior, the default isLevel Trigger behavior
EPOLLONESHOTSet one-shot behavior. After one event ispulled out, the fd is internally disabled
EPOLLRDNORMEquivalent toEPOLLIN
EPOLLRDBANDPriority data band can be read.
EPOLLWRNORMEquivalent toEPOLLOUT
EPOLLWRBANDPriority data may be written.
EPOLLMSGIgnored.
epoll.close()

Close the control file descriptor of the epoll object.

epoll.closed

True if the epoll object is closed.

epoll.fileno()

Return the file descriptor number of the control fd.

epoll.fromfd(fd)

Create an epoll object from a given file descriptor.

epoll.register(fd[,eventmask])

Register a fd descriptor with the epoll object.

epoll.modify(fd,eventmask)

Modify a registered file descriptor.

epoll.unregister(fd)

Remove a registered file descriptor from the epoll object.

epoll.poll(timeout=-1,maxevents=-1)

Wait for events. timeout in seconds (float)

Changed in version 3.5:The function is now retried with a recomputed timeout when interrupted bya signal, except if the signal handler raises an exception (seePEP 475 for the rationale), instead of raisingInterruptedError.

18.3.3. Polling Objects

Thepoll() system call, supported on most Unix systems, provides betterscalability for network servers that service many, many clients at the sametime.poll() scales better because the system call only requires listingthe file descriptors of interest, whileselect() builds a bitmap, turnson bits for the fds of interest, and then afterward the whole bitmap has to belinearly scanned again.select() is O(highest file descriptor), whilepoll() is O(number of file descriptors).

poll.register(fd[,eventmask])

Register a file descriptor with the polling object. Future calls to thepoll() method will then check whether the file descriptor has anypending I/O events.fd can be either an integer, or an object with afileno() method that returns an integer. File objectsimplementfileno(), so they can also be used as the argument.

eventmask is an optional bitmask describing the type of events you want tocheck for, and can be a combination of the constantsPOLLIN,POLLPRI, andPOLLOUT, described in the table below. If notspecified, the default value used will check for all 3 types of events.

ConstantMeaning
POLLINThere is data to read
POLLPRIThere is urgent data to read
POLLOUTReady for output: writing will not block
POLLERRError condition of some sort
POLLHUPHung up
POLLNVALInvalid request: descriptor not open

Registering a file descriptor that’s already registered is not an error, and hasthe same effect as registering the descriptor exactly once.

poll.modify(fd,eventmask)

Modifies an already registered fd. This has the same effect asregister(fd,eventmask). Attempting to modify a file descriptorthat was never registered causes anOSError exception with errnoENOENT to be raised.

poll.unregister(fd)

Remove a file descriptor being tracked by a polling object. Just like theregister() method,fd can be an integer or an object with afileno() method that returns an integer.

Attempting to remove a file descriptor that was never registered causes aKeyError exception to be raised.

poll.poll([timeout])

Polls the set of registered file descriptors, and returns a possibly-empty listcontaining(fd,event) 2-tuples for the descriptors that have events orerrors to report.fd is the file descriptor, andevent is a bitmask withbits set for the reported events for that descriptor —POLLIN forwaiting input,POLLOUT to indicate that the descriptor can be writtento, and so forth. An empty list indicates that the call timed out and no filedescriptors had any events to report. Iftimeout is given, it specifies thelength of time in milliseconds which the system will wait for events beforereturning. Iftimeout is omitted, negative, orNone, the call willblock until there is an event for this poll object.

Changed in version 3.5:The function is now retried with a recomputed timeout when interrupted bya signal, except if the signal handler raises an exception (seePEP 475 for the rationale), instead of raisingInterruptedError.

18.3.4. Kqueue Objects

kqueue.close()

Close the control file descriptor of the kqueue object.

kqueue.closed

True if the kqueue object is closed.

kqueue.fileno()

Return the file descriptor number of the control fd.

kqueue.fromfd(fd)

Create a kqueue object from a given file descriptor.

kqueue.control(changelist,max_events[,timeout=None]) → eventlist

Low level interface to kevent

  • changelist must be an iterable of kevent object orNone
  • max_events must be 0 or a positive integer
  • timeout in seconds (floats possible)

Changed in version 3.5:The function is now retried with a recomputed timeout when interrupted bya signal, except if the signal handler raises an exception (seePEP 475 for the rationale), instead of raisingInterruptedError.

18.3.5. Kevent Objects

https://www.freebsd.org/cgi/man.cgi?query=kqueue&sektion=2

kevent.ident

Value used to identify the event. The interpretation depends on the filterbut it’s usually the file descriptor. In the constructor ident can eitherbe an int or an object with afileno() method. keventstores the integer internally.

kevent.filter

Name of the kernel filter.

ConstantMeaning
KQ_FILTER_READTakes a descriptor and returns wheneverthere is data available to read
KQ_FILTER_WRITETakes a descriptor and returns wheneverthere is data available to write
KQ_FILTER_AIOAIO requests
KQ_FILTER_VNODEReturns when one or more of the requestedevents watched infflag occurs
KQ_FILTER_PROCWatch for events on a process id
KQ_FILTER_NETDEVWatch for events on a network device[not available on Mac OS X]
KQ_FILTER_SIGNALReturns whenever the watched signal isdelivered to the process
KQ_FILTER_TIMEREstablishes an arbitrary timer
kevent.flags

Filter action.

ConstantMeaning
KQ_EV_ADDAdds or modifies an event
KQ_EV_DELETERemoves an event from the queue
KQ_EV_ENABLEPermitscontrol() to returns the event
KQ_EV_DISABLEDisablesevent
KQ_EV_ONESHOTRemoves event after first occurrence
KQ_EV_CLEARReset the state after an event is retrieved
KQ_EV_SYSFLAGSinternal event
KQ_EV_FLAG1internal event
KQ_EV_EOFFilter specific EOF condition
KQ_EV_ERRORSee return values
kevent.fflags

Filter specific flags.

KQ_FILTER_READ andKQ_FILTER_WRITE filter flags:

ConstantMeaning
KQ_NOTE_LOWATlow water mark of a socket buffer

KQ_FILTER_VNODE filter flags:

ConstantMeaning
KQ_NOTE_DELETEunlink() was called
KQ_NOTE_WRITEa write occurred
KQ_NOTE_EXTENDthe file was extended
KQ_NOTE_ATTRIBan attribute was changed
KQ_NOTE_LINKthe link count has changed
KQ_NOTE_RENAMEthe file was renamed
KQ_NOTE_REVOKEaccess to the file was revoked

KQ_FILTER_PROC filter flags:

ConstantMeaning
KQ_NOTE_EXITthe process has exited
KQ_NOTE_FORKthe process has calledfork()
KQ_NOTE_EXECthe process has executed a new process
KQ_NOTE_PCTRLMASKinternal filter flag
KQ_NOTE_PDATAMASKinternal filter flag
KQ_NOTE_TRACKfollow a process acrossfork()
KQ_NOTE_CHILDreturned on the child process forNOTE_TRACK
KQ_NOTE_TRACKERRunable to attach to a child

KQ_FILTER_NETDEV filter flags (not available on Mac OS X):

ConstantMeaning
KQ_NOTE_LINKUPlink is up
KQ_NOTE_LINKDOWNlink is down
KQ_NOTE_LINKINVlink state is invalid
kevent.data

Filter specific data.

kevent.udata

User defined value.