When you cause a Pyramid application to start up in a console window, you'll see something much like this show up on the console:

$VENV/bin/pserve development.ini
Starting server in PID 16305.
Serving on http://localhost:6543
Serving on http://localhost:6543

This chapter explains what happens between the time you press the "Return" key on your keyboard after typing pserve development.ini and the time the lines Serving on http://localhost:6543 are output to your console.

The Startup Process

The easiest and best-documented way to start and serve a Pyramid application is to use the pserve command against a PasteDeploy .ini file. This uses the .ini file to infer settings and starts a server listening on a port. For the purposes of this discussion, we'll assume that you are using this command to run your Pyramid application.

Here's a high-level time-ordered overview of what happens when you press return after running pserve development.ini.

  1. The pserve command is invoked under your shell with the argument development.ini. As a result, Pyramid recognizes that it is meant to begin to run and serve an application using the information contained within the development.ini file.

  2. pserve passes the development.ini path to plaster which finds an available configuration loader that recognizes the ini format.

  3. plaster finds the plaster_pastedeploy library which binds the PasteDeploy library and returns a parser that can understand the format.

  4. The PasteDeploy finds a section named either [app:main], [pipeline:main], or [composite:main] in the .ini file. This section configures the WSGI webserver which serves your application. As such it is the ini section for your application and can be the source for many of your application's settings.

    If you're using a simple application (e.g., [app:main]), the application's paste.app_factory entry point will be named on the use= line within the section's configuration. If instead of a simple application, you're using a WSGI pipeline (e.g., a [pipeline:main] section), the application named on the "last" element will refer to your Pyramid application. If instead of a simple application or a pipeline, you're using a "composite" (e.g., [composite:main]), refer to the documentation for that particular composite to understand how to make it refer to your Pyramid application. In most cases, a Pyramid application built from a cookiecutter will have a single [app:main] section in it, and this will be the application served.

  5. The framework finds all logging related configuration in the .ini file and uses it to configure the Python standard library logging system for the application. See Logging Configuration for more information.

  6. The application's entry point, usually the entry point referenced on the above mentioned use= line, is the application's constructor. It is passed the key/value parameters in the application's .ini section. The constructor should return a router instance, which is a WSGI application.

    For Pyramid applications, the constructor is a function named main in the file within the package in which your application lives. If this function succeeds, it will return a Pyramid router instance. Here's the contents of an example module:

     1from pyramid.config import Configurator
     4def main(global_config, **settings):
     5    """ This function returns a Pyramid WSGI application.
     6    """
     7    with Configurator(settings=settings) as config:
     8        config.include('pyramid_jinja2')
     9        config.include('.routes')
    10        config.scan()
    11    return config.make_wsgi_app()

    Note that the constructor function accepts a global_config argument, which is a dictionary of key/value pairs mentioned in the [DEFAULT] section of an .ini file (if [DEFAULT] is present) and the executing script name __script__ like pserve, prequest or pshell. It also accepts a **settings argument, which collects another set of arbitrary key/value pairs. The arbitrary key/value pairs received by this function in **settings will be composed of all the key/value pairs that are present in the [app:main] section (except for the use= setting) when this function is called when you run pserve.

    Our generated development.ini file looks like so:

     2# app configuration
     7use = egg:myproject
     9pyramid.reload_templates = true
    10pyramid.debug_authorization = false
    11pyramid.debug_notfound = false
    12pyramid.debug_routematch = false
    13pyramid.default_locale_name = en
    14pyramid.includes =
    15    pyramid_debugtoolbar
    17# By default, the toolbar only appears for clients from IP addresses
    18# '' and '::1'.
    19# debugtoolbar.hosts = ::1
    22# wsgi server configuration
    26use = egg:waitress#main
    27listen = localhost:6543
    30# logging configuration
    35keys = root, myproject
    38keys = console
    41keys = generic
    44level = INFO
    45handlers = console
    48level = DEBUG
    49handlers =
    50qualname = myproject
    53class = StreamHandler
    54args = (sys.stderr,)
    55level = NOTSET
    56formatter = generic
    59format = %(asctime)s %(levelname)-5.5s [%(name)s:%(lineno)s][%(threadName)s] %(message)s

    In this case, the myproject.__init__:main function referred to by the entry point URI egg:myproject (see development.ini for more information about entry point URIs, and how they relate to callables) will receive the key/value pairs {pyramid.reload_templates = true, pyramid.debug_authorization = false, pyramid.debug_notfound = false, pyramid.debug_routematch = false, pyramid.default_locale_name = en, and pyramid.includes = pyramid_debugtoolbar}. See Environment Variables and .ini File Settings for the meanings of these keys.

  7. The main function begins by making a configurator. The dictionary captured via the **settings kwarg is passed to the Configurator constructor in its settings argument. The new configurator holds the application's settings and is able to commit any configuration declarations the settings contain.

    The settings dictionary contains all the options in the [app:main] section of our .ini file except the use option (which is internal to PasteDeploy) such as pyramid.reload_templates, pyramid.debug_authorization, etc. It is available for use in your code.

  8. The main function then calls various methods on the instance of the class Configurator created in the previous step. The intent of calling these methods is to populate an application registry, which represents the Pyramid configuration related to the application.

  9. The make_wsgi_app() method is called. The result is a router instance. The router is associated with the application registry implied by the configurator previously populated by other methods run against the Configurator. The router is a WSGI application.

  10. An ApplicationCreated event is emitted (see Using Events for more information about events).

  11. Assuming there were no errors, the main function in myproject returns the router instance created by pyramid.config.Configurator.make_wsgi_app() back to pserve. As far as pserve is concerned, it is "just another WSGI application".

  12. pserve starts the WSGI server defined within the [server:main] section. In our case, this is the Waitress server (use = egg:waitress#main), and it will listen on all interfaces on port 6543 for both IPv4 and IPv6 (listen = localhost:6543). The server code itself is what prints Serving on http://localhost:6543. The server serves the application, and the application is running, waiting to receive requests.

See also

Logging configuration is described in the Logging chapter. There, in Request Logging with Paste's TransLogger, you will also find an example of how to configure middleware to add pre-packaged functionality to your application.

Deployment Settings

Note that an augmented version of the values passed as **settings to the Configurator constructor is available in Pyramid view callable code as request.registry.settings. You can create objects you wish to access later from view code, and put them into the dictionary you pass to the configurator as settings. They will then be present in the request.registry.settings dictionary at application runtime.