Reactor-based Services
Here are services to run clients, servers and periodic services using the reactor.
If you want to run a server service, StreamServerEndpointService
defines a service that can wrap an arbitrary IStreamServerEndpoint
as an IService
. See also twisted.application.strports.service
for constructing one of these directly from a descriptive string.
Additionally, this module (dynamically) defines various Service subclasses that let you represent clients and servers in a Service hierarchy. Endpoints APIs should be preferred for stream server services, but since those APIs do not yet exist for clients or datagram services, many of these are still useful.
They are as follows:
TCPServer, TCPClient, UNIXServer, UNIXClient, SSLServer, SSLClient, UDPServer, UNIXDatagramServer, UNIXDatagramClient, MulticastServer
These classes take arbitrary arguments in their constructors and pass them straight on to their respective reactor.listenXXX or reactor.connectXXX calls.
For example, the following service starts a web server on port 8080: TCPServer(8080, server.Site(r)). See the documentation for the reactor.listen/connect* methods for more information.
Class |
|
A ClientService maintains a single outgoing connection to a client endpoint, reconnecting after a configurable timeout when a connection fails, either before or after connecting. |
Class |
|
Simple service.IService which starts and stops a twisted.internet.task.Cooperator . |
Class |
|
Undocumented |
Class |
|
Undocumented |
Class |
|
Undocumented |
Class |
|
A StreamServerEndpointService is an IService which runs a server on a listening port described by an IStreamServerEndpoint . |
Class |
|
Undocumented |
Class |
|
Undocumented |
Class |
|
Service to periodically call a function |
Class |
|
Undocumented |
Class |
|
Undocumented |
Class |
|
Undocumented |
Class |
|
Undocumented |
Class |
|
Undocumented |
Function | backoff |
A timeout policy for ClientService which computes an exponential backoff interval with configurable parameters. |
Class | _ |
No summary |
Class | _ |
No summary |
Class | _ |
State machine for maintaining a single outgoing connection to an endpoint. |
Class | _ |
A _DisconnectFactory is a proxy for IProtocolFactory that catches connectionLost notifications and relays them. |
Class | _ |
A proxy for a Protocol to provide connectionLost notification to a client connection service, in support of reconnecting when connections are lost. |
Class | _ |
Undocumented |
Function | _first |
Return the first element of a generator and exhaust it. |
Function | _maybe |
No summary |
Variable | _client |
Undocumented |
Variable | _default |
Undocumented |
Variable | _server |
Undocumented |
A timeout policy for ClientService
which computes an exponential backoff interval with configurable parameters.
Parameters | |
initialfloat | Delay for the first reconnection attempt (default 1.0s). |
maxfloat | Maximum number of seconds between connection attempts (default 60 seconds, or one minute). Note that this value is before jitter is applied, so the actual maximum possible delay is this value plus the maximum possible result of jitter(). |
factor:float | A multiplicative factor by which the delay grows on each failed reattempt. Default: 1.5. |
jitter:0-argument callable returning float | A 0-argument callable that introduces noise into the delay. By default, random.random, i.e. a pseudorandom floating-point value between zero and one. |
Returns | |
see ClientService.__init__ 's retryPolicy argument. | a 1-argument callable that, given an attempt count, returns a floating point number; the number of seconds to delay. |
Present Since | |
16.1.0 |
Return the first element of a generator and exhaust it.
MethodicalMachine.upon's collector argument takes a generator of output results. If the generator is exhausted, the later outputs aren't actually run.
Parameters | |
gen | Generator to extract values from |
Returns | |
The first element of the generator. |