[repo tests] add test case for RQLUniqueConstraint

[primary view] properly handle case where view specified in pvdc['vid'] isn't selectable

[primary view] don't need to pick arbitrary target, this is properly handled by the rtag

[schema] turn comment into docstring for RQLUniqueConstraint

[schema] ensure DISTINCT is displayed for constraint using it, this ease debugging

[windows distrib] trailing slashes in MANIFEST.in breaks windows install (sic)

[schema] mark CWUniqueTogetherConstraint and its relations as schema entity/relations

[schema] mark CWUniqueTogetherConstraint and its relations as schema entity/relations

[windows distrib] trailing slashes in MANIFEST.in breaks windows install (sic)

backport stable

merge stable heads

backport stable

Added tag cubicweb-debian-version-3.9.6-1 for changeset 7d2cab567735

Added tag cubicweb-version-3.9.6 for changeset d37428222a63

[migration] attempt to fix pre 3.6 migration

[migration] remove buggy assertion

[js] jqNode is more robust here wrt some node ids

[migration] .pop() is more readable

[form fields] option may be a 3-uple (3d element being dict containing widget options)

[migration] fix so that context modification are reflected on the currently executed migration script

3.9.6

[javascript] get rid of tabs

[javascript] fix test_ajax.js implementation In cubicweb.ajax.js, we maintain two lists (cw.loaded_src and cw.loaded_href) to keep track of already included CSS and JS. (see discussion about jQuery and $('foo').appendTo($head) in _loadAjaxHtmlHead.) Since all tests are executed in the "same page", those two variables need to be reinitialized before each test.

[migration] commit just after symetric has been renamed to symmetric (else later rollback may cancel this)

[doc] fix link to appendix

nicer error reporting for unique together constraints

removed print_exc

fix tests

unique_together integration in CW (alf+syt) * added in-DB representation of the constraint with a new entity type (CWUniqueTogetherConstraint) * the schema is now migrated on a sync_schema_props_perms(EType) call in the migration script * removed the migration commands provided in the first implementation

cleanup

[test] skip unittest_fti if we don't have the required setup...

[test] fix in 6203:d3dea5f84404 wasn't enough

[json controller] check rql is really specified

[test] during some tests (unittest_migrations), we should not register ClearGroupMap which is actually already there

i18n update

[c-c create] ask for database driver by default

[migration] when some CWGroup is added during migration, we should reset migration handler's group mapping cache and the set telling what have been synchronized (we may be able to add some new groups on perms syncing...)

remove debug print

add a small c-c shell script to detect cycle for a given relation

[test] fixes to follow recent changes

cleanups

[rset] add test for #1251252, actually fixed in rql; fix docstring and add comments in rset.related_entity to make things clearer

[querier test] remove some actually bad tests that were passing thanks to a sqlite bug Now, rql has been fixed to correctly detect this kind of query can't be answered correctly

[views] do not underline h1 titles for tabbedprimary views

[doc] improve doc thanks to user feedback

[schema deserial] fix loading of schema when mapping an entity type to an existing one To ensure rdef of the old entity type don't overwrite those already existing on the mapped entity type. Also rename some variables/arguments on the way to make things clearer.

[migration] more rename_entity_type fix * ensure repository caches and system tables are properly cleanup * delete relations from/to entities deleted through sql So we don't ends up with a corrupted db...

cleanups

[integrity] fix case where we can get two entities related while the cardinality is single on the object of an inlined relation

[migration] propose to open pdb on error during execution of rql/sql query

[migration] set ask_confirm=False by default on rqlexec

[c-c db-check] skip error while checking relation, useful when analyzing really broken database (after a migration failure for instance

[c-c db-check] add missing import

[doc] fix typo

merged unpushed changes

merged unpushed changes

fix name error by adding missing import

migration actions for unique_together constraints

remove buggy call to builtins._ on navigation content

[views] rename Edition to Modification and personal information to profile

[cw.utils] avoid spurious warning, ustrftime may be globally imported

typo

add missing msgstr

backport stable

[web/uicfg] better error reporting for reledit_ctrl

merge

merge

fix order of parameters in call to callfunc_every

[test/entity] fix test (following mtconverter 0.8+ plain text transform change)

[doc] add documentation on RealDatabaseConfiguration

[devtools] add url_publish() method on CubicWebTC to test a full publishing process from the url_resolver to the html generation

[devtools] make CubiwebTC.login() context-manager restore the testcase connection on exit

[devtools] restore ability to test on real databases

[migration] fix add_relation_type() for symmetric relations To check if the corresponding SQL table has to created, CWRelationAddOp tests that the tablename is not in the 'createdtables' session variable (i.e. the table hasn't been created before in the same transaction) and that len(rschema.rdefs) is not greater to 1, that is the relation already existed in the database before this transaction. In case of symmetric relations, len(rschema).rdefs will be at least 2 since rdefs will store both (subj, obj) and (obj, subj) relation definitions.

backport stable

Added tag cubicweb-debian-version-3.9.5-1 for changeset 12038ca95f0f

Added tag cubicweb-version-3.9.5 for changeset 0a1fce8ddc67

3.9.5

i18nupdate

[doc, adapters] drop deprecated code from examples

[doc] interesting stuff backported

[doc] clean ITreeAdapter documentation

cleanup

remove useless import

use is_instance in a number of places (esp. documentation) rather than the deprecated implements

[new components] fix dumb name error

remove some deprecated stuff about relation definition

backport stable

[book] more about has_<action>_permission and enhance a bit security definition section

update documentation to follow 6142:8bc6eac1fac1 changes. Try to make it better and move most doc with code on the way

cwconfig module already used by another autodocstring. Drop this currently useless file

more cwctl api update and fixes

[c-c create] argument inversed accidentally

merge

[session] cleanup hook / operation / entity edition api Operation api ~~~~~~~~~~~~~ * commit_event killed, recently introduced postcommit_event is enough and has a better name * kill SingleOperation class, it's a) currently never used b) superseeded by set_operation if needed. Entity edition api ~~~~~~~~~~~~~~~~~~ edited_attributes turned into a special object holding edition specific attributes: - attributes to be edited (simply mirrored in cw_attr_cache, actual values are there) - former _cw_skip_security set (cw_edited) and querier_pending_relations It has also been renamed to `cw_edited` on the way (it may also contains inlined relations) The entity dict interface has been deprecated. One should explicitly use either cw_attr_cache or cw_edited according to the need. Also, there is now a control that we don't try to hi-jack edited attributes once this has no more effect (eg modification have already been saved) At last, _cw_set_defaults/cw_check internal methods have been moved to this special object Hook api ~~~~~~~~ hook.entity_oldnewvalue function now moved to a method of cw_edited object.

[web api] unify 'contentnav' (VComponent) and 'boxes' registries as 'ctxcomponents' (CtxComponent)

[boxes] introduce new boxes system * separate box content generation from its layout * refactor css classes to allow moving boxes and still get consistent ui On the way to contentnavigation/boxes unification (in a later patch) * * * some fixes for the previous (default_new_boxes_system) patch * * * some fixes for the previous (default_new_boxes_system) patch

backport stable

update to lgc.clcommands 0.51 api

[css] make sure the toolbar is clickable because layered over title

[dataimport] refactor commitevery to gain readability

[win32] no os.getuid on windows

use logilab.common.date.ustrftime in json encoder to work around datetime.strftime limitation on dates < 1900

use iter_and_commit in example

[c-c check-mapping] fix dumb name error and add a warning about inlined crossed relation

[ms] more planning bug fixes * fix sourcesterms w/ constant: all occurences of the same constant should be there once one of them is * fix handling of constant and relation in expand_terms: * we should always consider system source there for constants * for relation, we should check its lhs and rhs are in selected terms, not that the relation is in linkedterms (this has no chance to be true) This fixes 3 tests that were either passing erroneously (syt should be punished for copy/paste output as test reulst without triple checking it) or were not optimal.

oops, missing part of 6125:628cf5213154 (mapping file checking)

[ms] fix two planner bugs: one occuring query such as X created_by U where X in a external source and U may come from an ldap source. The other being that when we've to merge input maps, we were modifying the same tree/solutions while a copy were needed. Also, ensure we add type restrictions, necessary for pyro source

cleanups

[ms, c-c] new command checking for consistency / potentian flaws and enhancements of mapping file of a multi-sources instance

[repository] enabled sources refactoring: to avoid error, we should always have all known source in repo.sources_by_uri and only enabled ones in repo.sources, so we still have access to the definition of temporarily disabled sources

[migration] enhance RqlForIterator so we can do 'for e in rqliter(...).entities()'