[components] remove old deprecation warning

[wf history] avoid title duplication

i18n tweaks

[book] small note about test/data/schema.py

backport stable

[test] reset uncommitable flag on ValidationError raised during automatic db population

[ctx component] has no wview method

[entity, repo] move prefill_entity_cache function back to repository due to usage of del_existing_rel_if_needed function which is also defined there

cleanup and 3.10 updates

[architecture] introduce adapter to ease transformation of errors before display to the end user. Use it for UniqueTogetherError first

[repo, entity] move entity cache initialization to a function, and call it before source.add_entity so it may be used in error handler or such

i18n update

[ui component] restore see_also component

[ui component] RelatedObjectsCtxComponent to superseede RelatedObjectsVComponent (still there for bw compat)

[c-c i18ncubicweb] fix 'non-contextual' relations

[i18n] uniformize footer capitalization

cleanup

[selector] more efficient rql_condition by simply retreiving COUNT()

[workflow] fix bw compat code

[baseviews] new cw.archive.[by_author|by_date] views, extracted from blog

[test/possible views] skip deprecated views

[form] unify form.render prototype to take a 'w' argument as other view/components render method

[ui] nicer ajax edit relation table (tags box for instance)

fix pre 3.10 bw compat

small typo on mustMach, handle initialvalue, handle empty userInput

[formwidgets] rewrite autocomplete cw widgets to get rid of div-style declaration API

[javascript] update to jQuery 1.4.2 and jQuery-ui 1.8.5 The most notable changes are : - A bug in jQuery 1.4.2 forces us to add 2 noop() methods on cw objects to handle event unbinding, see : http://dev.jquery.com/ticket/6184 for more details. - jquery.autocomplete.js is not maintained anymore as it now has a standard implementation in jquery ui. The downside is that the new autocomplete version needs more work to be used. cubicweb.widgets.js defines a cwautocomplete plugin which is just a (not so) thin wrapper around the std autocomplete plugin to keep backward compatibility and usage simplicty. The original autocomplete author wrote a nice migration guide : http://www.learningjquery.com/2010/06/autocomplete-migration-guide - If you're using jQuery.ajax directly, you might have surprises with arguments serialization for arrays. To get the old behaviour, you need to pass a `traditional: true` extra arguments to jQuery.ajax. Of course, it's transparent if you're using cubicweb higher level helpers such as loadxhtml, loadRemote, etc.

[tests] update windmill use cases from link corrections

[contextual components] missing layout for components in the toolbar

[selectors] make configuration_values selector works for startup/maintainance hook

[c-c i18ncube] proper command class name

[i18n cube] skip bw compat registries boxes and contentnavigation

backport stable

TestServerConfig needs default_base_url

backport stable

[test] we must consider default_base_url()

[js test] cleanup, move anonymous_logged handling to the test class (easier to control there

CtxComponent don't have wview

[test] missing main

backport stable

[test] fix ldap test when executed from certain hosts

[test] missing main

restore line dropped in 6424:f443a2b8a5c7 which is necessary to have proper base-url on test server config (used for ms tests at least)

typo

[breadcrumb] infinite loop safety belt, bw compatible by hi-jacking 'recurs' argument

[login] fix some previously introduced login box css breakage

[components] refactor main template header: make it much more flexible by using CtxComponent * rename userlink class (not private) with somewhat bw compat * make some cosmetic adjustements on log-in/registration components * update to component api, though this may breaks some sub-classes (for application customizing the breadcrumbs component for instance)

[schema / sources] store data sources as cubicweb entities this implies several changes: * new CWSource / CWSourceHostConfig entity types * only the system sources and default admin login/password stored in sources file (other stuff will be ignored) * on startup, get sources definition from the database * every entities have a cw_source relation * a facet allow filtering

[hook/operation] nicer api to achieve same result as set_operation, as described in #1253630

backport stable

[devtools] refactor http server initialization in a much saner way Use a new ApptestConfiguration to group server paramater. Fix base-url mangling between main site and data ressources.

[pre 3.10 compat] if CtxComponent as a call() method, emit a deprecation warning but use it. Also ensure default value on class for visible and order, and define wview for bw compat as well

[reledit] do not reload on the parent node

[js/edition] kill deprecated (& dead) reledit code

[views/wdoc] do not crash on malconstructed stuff

typo

[views/authentication] fix http auth regression (no message)

[test] fixes

fix bad merge

cleanups

[test] skip test when dbhelper has no underlying dbapi module available

backport stable

[test] cleanup / unittest2 api update

[test] don't add unnecessary dependency to blog

[test] fix tests broken by transaction behaviour on Unauthorized/ValidationError (no rollback but connection marked as non-commitable)

backport stable

[testlib] fix mail monkey patching, former patch was applied on stable instead of default

[primary view] default to outofcontext view for entities displayed in boxes

xml-escape <script> tags generated by HTMLHead

fix some issues in DotGraphView, and enhance reusability

backport stable

[testlib] Fix mail monkeypatch in testlib ea26eb5fd388 in stable introduce monkeypatching of SendMailOp.commit_event 8bc6eac1fac1 in default rename SendMailOp.commit_event to postcommit_event d7f5d873e1b8 merge stable in default without handling the rename.

[selector] fix iterable detection

backport stable

restore vital block of code (gone in 6384:89d5b339ebdd)

[debian] constrain pyro version in debian packaging

[test] send mail synchronously during tests

[book] improve documentation lisibility for cwconfig chapter and fix some references

[login ui] cleanup and somewhat bw compat

[selectors] move selectors according to their type, add missing ones to documentation

[forms] remove dead code and simplify inlineviews creation remove AutomaticEntityForm.should_inline_relation_form which is not used anywhere and comes from a time where uicfg did not exist. remove test for "force_<rtype>_display" presence in req.form to determine whether or not the inline creation form should be displayed.

include fcayre's patch #1060507 to get extra rql query parameters from form content in rgx_action

[forms] catching KeyError closes #961831: migration failure when defining form fields

[authentication] force retriever implementor to think about it

[repo] add a find_users method for under-privileged accounts

refactor login box & form to enable easy pluggability * vregistry.selectable: get all selectable object of fixed oid with given context * template headeri, logbox, logform: reorganize a bit the structure

[js] add cubes namespace

new script cwuser_ldap2system used to migrate ldap users to system source based on work by adim.

[vreg] fix bad copy and paste introduced in 6347:ef47a2100c66

[repo transaction] fix rollback behaviour as discussed on the mailing-list: instead of rollbacking automatically on Unauthorized/ValidationError, mark the transaction as uncommitable and disallow commiting

#1256570: remove __method, it is not used in any of the cubes from logilab and encourages bad design by using Redirect (eg ui specific) from entities.py

Add a fsimport context manage to localy enable fsimporting.

[test] improve tag for server/test/unittest_storage.py Use class level tag.

[repo threads] Add several safety when looking for a callable name. We lookup __name__ attribute, then func_name attribute. object repr is a last fallback

[facets] support for `no_relation` on RelationFacet e.g. to filter en entities *without* a given relation as well as with a particular relation. Proposed or not by default according to relation cardinality. Also, try to guess i18nable properly from the schema by default.

[R] use dbhelper.binary_value to process passwords and other Bytes fields (instead of dbapi_module.Binary or hardcoding buffer()). This requires (esp. in serverctl) version 1.3.1 of logilab.database. Dependencies updated.

[3.10 migration] fix migration of cwproperties related to components/boxes

[schema] introduce some new sets categorizing entity/relation types and benefits from them where possible

backport stable

[schema] introduce new WORKFLOW_RTYPES set and use it to build SYSTEM_RTYPES/DONT_CHECK_RTYPES_ON_ADD/DONT_CHECK_RTYPES_ON_DEL sets

[rset] support for delitem on repeat list (may be necessary in pyro source)

merged timeline changes into default

timeline view minor enhancements * support dates < 1900 by using logilab.common.date.ustrftime * make the timeline a public member of the TimeLineWidget * let buildWidget return the created widget, so that we can manipulate it later

fix bad merge

backport stable

[test] more update to unittest2 api

[repo] properly mark object as saved once added to its source (code much probably gone during a merge...)

fix merge, some buffers weren't saved...

backport stable

[test] must now commit file creation since the later ValidationError trigger a rollback

cleanup and micro-optimization

[web action] has_editable_relations should not filter out final relations, fix regression introduced in 6358:22c95c5ef12d

[c-c] fix RuntimeError: 'maximum recursion depth exceeded while calling a Python object' we get when creating/upgrading/shelling an instance: hasattr() call __getattribute__, creating an infinite recursion error catched by the interpretor. Avoid this by testing the method is available on the class instead of the instance

[transaction] to avoid potential db corruption, we should rollback systematically in case of ValidationError

[selector] cleanup

[sync schema] take care rdef may not be final, in which case we want to use type of eid attribute

[web action] has_permission_update checked implicitly by has_editable_relation, don't check it twice

[muledit] form may contains subforms with nothing editable that will be s kipped, so ensure we choose one with something to display when rendering table headers

[c-c i18n] namespace/line wrap cleanup + dynamically compute available language + make it works in installed mode

[i18n] no more necessary, those msgids are no more generated