backport stable

[db-check] fix sql to fix bad eid referenced by inlined relation

use 3.6 attributes

typo

this has probably be killed by a bad merge...

[notification] use super_session to select the recipients finder

backport stable branch

Added tag cubicweb-debian-version-3.5.2-1 for changeset 75cc4aa76fb1

Added tag cubicweb-version-3.5.2 for changeset 1f0aa3cd5af2

preparing 3.5.2

[i18n] updated catalogs

fix uid generation for py >= 2.5

update to 3.5 wf api

[magicsearch] update to match new i18n API: config.translations[lang] now returns a couple of function

[tests] fix inline-creation/edition unit tests

cleanup, reorganize

[css] backported download box styles from forge

[forms] do similar refactoring for inline edition than for inline creation

merge

refactor to ease overriding

[inline forms] propagate kwargs to render_form

[forms] propagate parameters added to autoform method

[schema sync] fix so that new cardinality are considered when both inlined and cardinality change in the same transaction

a case for tidy : for your consideration

[uicfg] automatically set generated for the opposite of inlined relation as well

since we have the transition entity, give it directly

merge

this is actually a role, fix argument name

fix vocab param to avoid deprecation warning

fix case insensitive selector for entity creation

[selector] check read perm if action is add/update/delete

don't want this

[validatecontroller] allow additional args to be passed to the js callback

[javascript] make inlinedform-added event bindable $(form).trigger('inlinedform-added') was not usable since form had no existence outside the inner callback. Use the general CubicWeb object to trigger the event so it's easy to bind a callback wherever it's needed.

[forms] take a look at inlined form to guess whether the form requires multipart or not

[form] simply skip unknown argument given to form constructor, don't fail

fix argument given to changestate form selection

can't use anymore make_uid to generate a valid js var name

remove debug raise

merge

statuschange form view only works w/ one entity, fix selector accordingly

fix sort of fields vocabulary: should consider option groups

[widgets] change make_uid() seed to make slider ids xhtml-valid

introduced to ease overriding for fancy stuff...

cleaner code

don't propose inlined form if user can't edit the entity

cleanup, smarter sort control to avoid it when not necessary

facets enhancement: can now set label_vid to use a view instead of a simple attribute as label

fix/enhance i18n context usage for inlined forms

don't show label if no label given

fix sys import

merge

merge

[D] book: define a workflow

cleanup

more in the way of subclassability & args passing all the way down do not set cookie if the tree is throw-away

merge

documentation : fixed docstring Added missing import in the provided data importation example script.

backport 3.5

backport old head

STABLE IS NOW 3.5

Added tag cubicweb-debian-version-3.5.1-1 for changeset f476cecd4690

Added tag cubicweb-version-3.5.1 for changeset 77ed72f3c260

3.5.1

backport stable

i18nupdate

we may have some entity forms with nothing to edit for an entity, no RequestError in that case

fix generated label for object relation

[wf graph] translate group names as well

[wf] uicfg tweaks

[migration] fix initial_state migration: at this point, S state_of ET is no more true (but we don't need it...)

[ui] translate state/tr names in wf image

cleanup

merge

remove 2.5 construct

missing import

make it easier to subclass

backport from 3.5

merge

Added tag cubicweb-debian-version-3.5.0-1 for changeset 7e5d0ae8d202

Added tag cubicweb-version-3.5.0 for changeset 70c0dd1c3b7d

do not wipe a good value if input is bad and cancelled

backport 3.5

backport stable

Added tag cubicweb-debian-version-3.4.11-1 for changeset 635a25031f4a

Added tag cubicweb-version-3.4.11 for changeset f3d2adf48332

bump

skip SYSTEM_RTYPES by default in schema view

tweak catalog generation by adding a new SYSTEM_RTYPES set to schema

remove debug print

fix .po generation: ensure we get all msgids when building a cube catalog

ensure pgettext return unicode as well

more doc

i18n roundup

[i18n] do not add context information for metadata relations

ensure bw compat for relation traduction, add context to _object msg id

patch gettext for compat w/ catalog generated by cw < 3.5

merge

fix see_also permissions

[selectors] score_entity selector now ensures an int score is returned

[gettext] provide GNU contexts to avoid translations ambiguities

gettext] provide a more up-to-date version of gettext.py taken from python2.6 + patched to defined pgettext (i.e. contextual translations)

typo

nicer add urls

set strict when checking if edit action should be shown

warning -> critical, also allow goa to work without vobject

prevent some command providers to stop using cubicweb-ctl (current case : goactl when vobject is not there)

3.5

backport default fix

we must now explicitly call rebuild_infered_relation

test update, no more a FakeVReg, needs to hack it

backport stable

i18n update

uicfg tweaks for wf

remove annoying sqlite, not cw, test

[jsoncontroller] bugfix: update to match new validate_form API

ensure unittest_pdf passes

update the documentation

[D] add entity inheritance example

[C] uicfg: Change CWCache as system entites

cleanup

should not have been commited

email hooks tests

tests update

those items are in the add menu, we don't want 'add' in their label

refactor to allow explicit card integrity check from client code (eg email hooks for instance)

fix rqlannotation bug w/ identity relation in subscope

fix cardinality integrity and security when setting use_email/primary_email

hide bookmarked_by relation from addrelated menu, we've a specific box for bookmarks

stop setting sort to false by default on relation field and explain why

cleanup

since unrelated_rql now consider security, we may consider returning relation having local role in some cases

unrelated_rql now return rql *and* args

missing import

unrelated_rql now considers relation's add perm

refactor so that rql rewriter may be used outside the server. Enhance it to be usable for RRQLExpression as well

[ms] fix planning w/ query using identity

[querier] fix sql generated w/ NOT relation and shared variable: ensure variable's table is in parent select'scope

[cleanup] remove unused imports

add note

fix when Interval is coming from date substraction

remerge

merge

[controllers] allow onsuccess / onfailure callback to be passed to validateform For instance : validateform?__onsuccess=wp.addAndSelectHall will call the addAndSelectHall function on success rather than redirecting to a specific URL.

[utils] provide a class to extend the default simplejson encoder to be able to dump standard yams attribute types

backport 3.5

test update

[actions] refactor: extract actual_actions from fill_menu to work with table filter form

test fix

should now use _test_action, not bare action.url()

facilities to test new submenu actions

comment out deprecated test

merge

was lost in some merge

[i18ncube] access to appearsin_addmenu config through uicfg directly

cleanup, futur warning fixes :)

edit box refactoring to gain more control by using actions

[3.5 migration] avoid removing defined transitions...

cleanup

[wf graph] fix display of condition

fix workflow graph views and tabs for the new wf system, remove useless view and action (actually useless since introduction of tabbed entity type view

[3.5 migration] missing identical_to/canonical -> prefered_form migration

[migration] don't ask confirm here

displaytime attribute should not have been removed from there

[forms] avoid to create superfluous add_new_relation_link in create form

merge

allow user to give ?tab=tabname and be set to the right tab (part of #345399)

merge

return self, not None, when no prefered form

tests update

fix entity classes inheritance (repair cw.test.unittest_entity)

#370578: change EmailAddress identical_to/canonical to prefered_form

merge

as shell, also works fine/linux

HOME is spelled USERPROFILE on windows

#345461: drop support for cube_eid / direct bug submission

backport 3.5

backport stable

[repo] take care of inlined relation in session.[add|delete]_relation

[repo] set entity.rset & co when needed

remove extra paren

[controllers] fix bad if/elif switch in handle_attribute

should change process'user once pid file has been written

windows doesn't like ':' in filesc

enough annoying for a 3.4.10...

another friend

merge

[c-c] fix 3.4.9 name error :'(

Added tag cubicweb-debian-version-3.4.9-1 for changeset 24ea70f19a48

Added tag cubicweb-version-3.4.9 for changeset e916f1e856c8

backport 3.5

backport stable branch

3.4.9

cleanup

fix daemonize: set umask, move to /, consider uid

[c-c] fix start/restart commands

from 3.5 branch; XXX the test depends on an eid value

merge

small note on fetch_unrelated_order ... fetch_order needs clarification

hopefully correctly fix the breadcrumbs escaping

Added tag cubicweb-debian-version-3.4.8-1 for changeset 97273eeaaead

Added tag cubicweb-version-3.4.8 for changeset ba43e084e884

missing escape

[vreg] we must ensure that parent classes are initialiazed first

[repo] oops, glob_update_entity takes a list of edited attributes has third argument

3.4.8

merge

[tests] inheritance_precedence in url rewrite

[D] add example about site properties configuration

properly escape site-title

[views] remove parameter displaytime from final view

nicer fix

backport 3.5

merge

[notification] should get actual session here

merge

Added tag cubicweb-debian-version-3.4.7-1 for changeset ebb92e62eb04

Added tag cubicweb-version-3.4.7 for changeset 0e549b299f0b

[editcontrollers] handle_attribute: test for value presence before trying to convert it to int/float

very dubious but observed : os.system works on linux, subprocess.call on windows, no other way around ?

oops, key can be non str

merge

fix user login for windows

close file handler leak

note about this test under windows

remove bogus assertion

note about licence, fix copyright, fix case of unicode argument

close file handler leak

use proper locks instead of hasardous time.sleep()s

have a better make_uid function, esp. useful for win32 as collisions are frequents with the old one

avoid open file handler leak

Bugfix: message was not written in english

merge

[javascript ie6] make "strptime" function work in ie6

ii8nupdate

update test now that etype_class systematically copy registry classes

merge

add pyro

fix so that we ensure issubclass(etype_class('SubNote'), etype_class('Note'))

backport stable