view @ 6028:1893da1d5213

userid lookup caches: use 1 on-disk cache file, update cache rather than rebuild Before this, we maintained one cache file per attribute (e.g. name2id, openid2id, ...) - the related code did multiple passes over all user profiles to rebuild these cache files. Now doing a one-pass rebuild, writing all attribute -> userid mappings into one on-disk cache file called "lookup". Additionally to "name" and "openids", support fast lookup for "email" and "jid" also. On profile save, we use to just kill the cache and let it rebuild. Now the cache is read, updated and written back (which is much less expensive for wikis with more than a few users). Did some refactoring also, reducing duplication, breaking down the code into smaller functions / methods.
author Thomas Waldmann <tw AT waldmann-edv DOT de>
date Wed, 12 Feb 2014 18:22:10 +0100
parents f105598176f5
line wrap: on
line source

# MoinMoin - Distutils distribution files
# Copyright (c) 2001, 2002 by Juergen Hermann <>
# All rights reserved, see COPYING for details.

# additional files not known by
include README setup.cfg moin.spec
include wikiserverlogging.conf

# documentation and licenses
recursive-include docs *

# include stuff for translators
recursive-include   MoinMoin/i18n *

# include static htdocs
recursive-include   MoinMoin/web/static/htdocs *

# include non-py stuff from werkzeug
recursive-include   MoinMoin/support/werkzeug/debug *

# contrib stuff
recursive-include   contrib *

# tests stuff
recursive-include   tests *
global-include */_tests/*

# "wiki" directory
recursive-include   wiki *

# omit revision control, python bytecode
global-exclude *.pyc
global-exclude *.pyo
global-exclude */CVS/*
global-exclude */.cvsignore

# we distribute a uncompressed version, no need for that:
global-exclude underlay.tar
global-exclude README.underlay