pyramid-dogpile-cache2 1.1.2

Integrates dogpile.cache for Pyramid

pyramid_dogpile_cache2

Small dogpile.cache configuration and access package. It is inspired by pyramid_dogpile_cache, which we found unusable since it insists on configuring the cache regions in its get_region() API – but if you want to use the @cache_on_arguments decorator, that is at import time, where no configuration exists yet. Our package wants to perform the configuration during the WSGI application setup instead.

This package is compatible with Python version >=3.6.

Usage

The package offers only one API function; it returns a dogpile.cache CacheRegion:

from pyramid_dogpile_cache import get_region
region = get_region('foo')

As said above, this is safe to call at import time, so you can go on like this:

@region.cache_on_arguments()
def expensive_function(one, two, three):
    # compute stuff

Setup / Pyramid

Include the package, either in code:

config = Configurator(...)
config.include('pyramid_dogpile_cache2')

or in the ini file:

pyramid.includes = pyramid_dogpile_cache2

Setup / Paste

For non-Pyramid WSGI applications that use a paste.ini file, you need to call:

def my_paste_app_factory(global_conf, **local_conf):
    pyramid_dogpile_cache2.configure_dogpile_cache(local_conf)
    return my_wsgi_callable

Settings

The settings support of pyramid_dogpile_cache unfortunately is quite incomplete (e.g. it does not even convert expiration_time to int). The support of this packages is a little better, but still very much incomplete: we support the in-memory and memcached backends (pylibmc to be precise), and only the same backend and configuration for all cache regions.

The following settings are supported:

dogpile_cache.regions

A list of region names that should be configured (separated by either spaces or commas).

dogpile_cache.backend

The default backend for cache regions (e.g. 'dogpile.cache.memory', dogpile.cache.pylibmc, etc.).

dogpile_cache.REGION.backend

Backend for the given region.

dogpile_cache.expiration_time

The default expiration time. Can be overridden for individual regions (in seconds). Omit to set no expiration.

dogpile_cache.REGION.expiration_time

The expiration time for the given cache region (in seconds). If omitted, uses the global expiration_time setting.

dogpile_cache.arguments.*

Defaults for backend arguments. Can be overridden for individual regions.

dogpile_cache.REGION.arguments.*

Backend arguments for the given cache region.

Backend arguments work only for strings, thus we support some custom treatment:

dogpile_cache.pylibmc_url

A list of memcached servers, separated by ;.

dogpile_cache.pylibmc_behavior.*

Set pylibmc behaviours, see coerce_memached_behaviors for which subkeys are supported.

Note: As opposed to pyramid_dogpile_cache we don’t support overriding the key_generator or key_mangler functions yet; we preconfigure them with enhanced versions of dogpile.cache that support non-ascii function arguments and generating cache keys for methods that include the class name.

Developing pyramid_dogpile_cache2

Author:

Zeit Online

PyPI page:

https://pypi.python.org/pypi/pyramid_dogpile_cache2

Issues:

report by e-mail

Source code:

https://github.com/zeitonline/pyramid_dogpile_cache2

Current change log:

https://github.com/zeitonline/pyramid_dogpile_cache2/blob/master/CHANGES.txt

Change log for pyramid_dogpile_cache2

1.1.2 (2022-07-15)

• Clear internal _actual_backend during configure, to improve test isolation

1.1.1 (2020-09-11)

• Ignore empty pylibmc_url setting

1.1.0 (2020-07-28)

• Drop Python-2 compatibility, update to dogpile.cache>=1.0

1.0.6 (2019-11-08)

• Allow configuring no expiration time at all

1.0.5 (2018-12-14)

• Clarify Python-3 compatiblity (it’s >=3.4, not _just_ 3.4)

1.0.4 (2018-11-21)

• Support caching functions with type annotations on Python 3 (see PR #5).

1.0.3 (2017-02-14)

• Fix packaging issue (see PR #2).

1.0.2 (2016-08-01)

• Update API compatibility to dogpile.cache-0.6.0.

1.0.1 (2016-01-20)

• Actually include the class name in the cache key for methods.

1.0.0 (2016-01-19)

• Initial release.