[repository] add a ``gc_stats`` method and 2 more cache size for ``stats`` (closes #2179735) This method returns memory informations in a dictionnary (same informations as the content displayed by the GCView).

[test] fix test broken by RichString handling changes in yams #default

[test] fix test broken by EmailAddress read permissions change

[test] fix test broken by renaming of module containing the cwxml parser and handling of vid in urls

[test] fix test broken by selectors/predicates changes coming with the move of registry to lgc

[repo test] fix test broken by recent changes on EmailAddress read permissions

backport stable

[server test] fix test broken by recent changes in lgdb

backport stable

[schema sync hooks] fix format string

[event bus] event has been renamed to publish

[schema view] properly display inheritance (closes #870073).

[workflow view] Display a non-localized version of state/transition names (closes #1912537).

[repo] Add a publish/subscribe mechanism for inter-instance communication using zmq Each repo can have a publishing and any number of subscribing sockets whose addresses are specified in the instance's configuration. An application or cube can subscribe to some 'topics', and give a callback that gets called when a message matching that topic is received. As a proof of concept, this introduces a hook to clean up the caches associated with the repository when an entity is deleted. A subscription is added using Repository::zmq::add_subscription; the callback receives a list representing the received multi-part message as argument (the first element of the message is its topic).

[migration] fire hooks in postcreate by default (closes #1417110) also turn usage of hooks_control to relevent session's method

[migration] `add_relation_type` warns and add missing definitions on already existing rtypes (closes #2121979)

[i18n] fix translations of login actions in french (closes #1546397)

([config] add file encoding information to python files (closes #1942655)

[primary view] delete PrimaryView.skip_none attribute (closes #2093250) it was already broken anyway, since introduction of reledit (a while ago)

[entities system table] change asource and source column size (closes #1951819)

[view] fix error messages of NotImplementedError

kill dead code in Workflow view This looks like it's been missed in r5224:34e669b6fd95.

[bw compat] add compatibility for import of former VRegistry

[views/attribute] role parameter must default to "subject" (closes #2175427)

[windows service] Copied the 3 lines from etwist/twctl into etwist/service in order to also clean the cache in Windows service mode. Could be factorised with etwist/server.run

[entity] fix crash when using cw_instantiate with empty related entity list; closes #2094462

[pylint] correctly detect uiprops module when it's not in a package

backport https-deny-anonymous option type fix done in default

Added tag cubicweb-debian-version-3.14.3-1 for changeset 4d0f5d18e8a0

Added tag cubicweb-version-3.14.3 for changeset 60efdbb45520

3.14.3

[web/formwidgets] make default_size a tunable, fix code layout (closes #2165936)

[hooks/syncschema] do not crash if the schema did not learn (yet) some new META rtype (closes #2171313)

[js] better support for no js in tabs (closes #2162092)

[vreg] move base registry implementation to logilab.common. Closes #1916014 A new logilab.common.registry module has been created with content from * cw.vreg (the whole things that was in there) * cw.appobject (base selectors and all). In the process, we've done some renaming: * former selector functions are now known as "predicate", though you still use predicates to build an object'selector * hence `objectify_selector` decorator is now `objectify_predicate` * the top level registry is now `RegistryStore` (was `VRegistry`) Also there is no more need for the @lltrace decorator. On the CubicWeb side, the `selectors` module has been renamed to `predicates`. There should be full backward compat with proper deprecation warnings.

[datafeed] when checking for deletion using cwentity xml parser, use vid=xml to limit resources consumption

[repository] ldap-feed source. Closes #2086984 datafeed based source which copy a subtree of the ldap directory into the system database. Authentication still go through ldap though. Pros: * don't need temporary tables and such for multi-sources RQL queries execution * much more flexible to enhance / configure behaviour (you simply have to replace the parser) * run better when ldap isn't reachable Cons: * no more 'on the fly' discovery of users (though a user authenticating itself will be automatically added if it doesn't exist in the db yet) * synchronization may be heavy if there are a lot of users A new cw.server.ldaputils containing code in common between former ldapuser and new ldapfeed sources has been introduced. Also ldapuser source now uses url instead of custom host/protocol option so it looks like a datafeed source (could be improved).

[datafeed] set delete-entities=yes is now safer, by checking each entity actually seems deleted. Closes #2165381 Parsers now have to implement the `is_deleted(extid, etype, eid)` method that should return a boolean telling if the given entity exists in the external source or not.

[config, migration] fix https-deny-anonymous option type

[js, i18n] translated strings should use " not ' else they are discarded by xgettext. Closes #1892489

[book] Fix typos in rql datetime functions documentation

[search box] allow to specify domid in select arguments. Closes #2162166 needed when one want to display search box twice in a page (eg logilabfr)

[config] fix rql-cache-size initial values (closes #2162160)

[migractions] fix comment to avoid ambiguity. Closes #2015260

[storage] BFSS now create read only file (closes #2151672) In the process binary.to_file now takes a file instead of a file-path. As storage was currently the only user it should not impact anyone.

backport stable

[schema synchronization] don't use sql for TODAY/NOW in keyword arg. Closes #2154815 Also consider TZDatetime on the way

backport oldstable

[req cookie] fix remove_cookie expires which was leading to expires computed to 0 in set_cookie and the Cookie class interpret that has no expires. Closes #2154654

[testlib, postgres] take care, _repo is None when backuping postgres test db, crashing database generation

[base views] make it easier to override EntityStartupView behaviour when the rset is empty; closes #2126364

[excanvas js] missing comma

backport stable

[datafeed] Don't pretend all columns are sortable in datafeed sources import log view. Closes #2146387

[facets] Make hiding facet body optional (closes #2105862)

[manage/index view] fix sources management view appearing in index view although it should not; closes #2126335

[book] add doc about request, session and connection management you need to know when writing tests

[test] test/demonstrate relation permissions overriding

[datafeed] Don't pretend all columns are sortable in datafeed sources import log view. Closes #2146387

[facets] Make hiding facet body optional (closes #2105862)

gmap: fix regression from 151058945234 Send zoomlevel as an int, and move closing brace to the right place.

gmap: fix regression from 151058945234 Send zoomlevel as an int, and move closing brace to the right place.

[deprecation] add cw version number to deprecation warning

[test] fix test broken by 8158:2ee254e74382 and add a test for that change

[doc] Add a small docstring to RQLExpression constructor

backport stable

[schema] restrictive email address read permission. Closes #2148141 You may have to update your application schema if you were based on previous settings!

[tableview] drop useless code: cellvid systematically given by RsetTableView, according to its .[non]finalview attribute

[tableviews] use ``select_or_none`` instead of ``try: select() except:`` select_or_none is here to avoid multiplying try: except: context in the code.

[bw compat] json_request setter is necessary for some cubes which has json_request (eg preview)

gmap: set center and zoomlevel automatically (closes #2144255) Instead of hardcoding the default zoomlevel and setting the center to the average of all markers, let gmap compute them so all markers fit in the map.

[views] calendar id in CalendarView must be unique (closes #2143864)

[debian] cubicweb works with python 2.7 (which become the standard on some distros). Closes #2142050

[multi-sources] call doexec rather than sqlexec since we do not want to call process_result (as there aren't any and this breaks with lgdb default)

[sources] hide some attributes from source edition form. Closes #2144160

[https] Allow configuration of anonymous access on https (closes #1971992) For now any anonymous access in denied on secured connection (https-url). This patch introduce a configuration option to control this behaviour. New instance will allow it. Old instance should stay as is during migration.

[bfss] convert root BFSS directory to str in constructor (closes #2137793) otherwise you can get crashed with AssertionError: Binary objects must use raw strings, not <type 'unicode'>

[rql, security] fix buggy rqlstr cleanup when some introduced snippet create an unresolvable query. Closes #2142041

[ldap test] test actual ldap authentication instead of monkey-patching also drop usage of not-anymore-necessary constants.

[book] Add some details about usage of EXISTS

backport stable

Added tag cubicweb-debian-version-3.14.2-1 for changeset 17ebd836cee3

Added tag cubicweb-version-3.14.2 for changeset 049a3819f03d

3.14.2

[i18n] fix french for 'action menu'

[server] use lgd.process_cursor to optimize processing of large resultsets update lgdb dependency to ensure we've a fix necessary for proper behaviour of the previous changeset

[css] fix typo in uiprops/css

backport stable

[sources management] fix crash mapping tab because chain() is not indexable. Closes #2124077

[css] use listingHeaderBgColor in old css this also fix inconsistency between background image / bg color

add instance to not implemented error

[repo cache] fix cache clearing for an eid: keys are now always a tuple

missing import statement

fix potential problems when BFSS uses a Windows SMB share (closes #2131435) files on windows network share must be read / written by chunks of reasonable size or you get some unusual os level errors.

[views] (in/outof)context views now use directly dc_(long)_title() (closes #2110235)

[table] use cell text content when cubicweb:sortvalue is not defined (closes #2093183)

[controllers] deprecate JSonController and implement AjaxController / ajax-func registry (closes #2110265)

[rql2sql] None for attributes in kwargs generate IS NULL, so should be considered in sql cache key. Closes #2116693

[tableview] cubicweb:sortvalue should not have 'json:' in its value anymore. This breaks old tables.

[schema] React to yams improvement of metadata attribute handling. * Change the way RichString is deprecated. * Add documentation about metadata attribute

stable is 3.14.X

oldstable is 3.13.X

Added tag cubicweb-debian-version-3.14.1-1 for changeset 6928210da4fc

Added tag cubicweb-version-3.14.1 for changeset 793d2d327b3e

3.14.1

backport stable

Added tag cubicweb-debian-version-3.13.10-1 for changeset 2eab9a5a6bf8

Added tag cubicweb-version-3.13.10 for changeset 2ad4e5173c73

[ajax] make navigation component appear both above and below the view (closes #2110266)

3.13.10

Link to python docs about strftime instead of pointing at "man" (closes #2106415) Not ideal, but better than pointing all users at Unix command-line tools.

[test] reindent

[rqlrewrite] fix comment

[server test] Fix tests regression from changeset 7980b36fb1aa

Tweak message to clarify it gets the pagination back http://www.cubicweb.org/task/2106487

fix tests broken by lgdb update

[migration] Ignore IOError from readline.write_history_file (closes #2106621) We already ignore them when reading the histfile, no reason writing it back should be any different.

[migration] make remove_cube('foo') not fail if foo is already removed. Closes #2106532

[cwconfig] absolute import (closes #2091068)

[schema sync] Update repo._type_source_cache when renaming an entity (closes #2094470) When an entity type is renamed as part of a migration, it leaves stale entries in repo._type_source_cache, which makes subsequent operations on those entities fail (they look for the old entity name instead of the new one).