[json controler] extraargs in js_component contains unicode keys which may not be given as argument names

[footer template] refactor to ease addition of some content into the footer

[boxes] add id of the rset to generated dom id of RsetBox to lower chance of id conflict (object id may be reused later)

[doc] fix bad class ref

oops, forgotten to save buffer...

backport stable

[unittest2] use unittest2 module fixture api

[server/test] raise SkipTest from setUpModule instead of setUp( unittest2 behaviour) + kill some deprecation warnings

[daemon] make process umask configurable

properly tell we've started an instance (close #1391262)

[ctxcomponents] RQLCtxComponent: refactor init_rendering/render_body to ease overriding init_rendering now simply execute the query to get result set to be displayed, and store it in .cw_rset attribute. Links to be displayed by default are now computed in render_body. So you may now simply customise what to be displayed by overriding render_body.

[tests] use TextTestRunner from unittest, no more need to use SkipAwareTextRunner from lgc.testlib (which doesn't exist anymore in recent releases

[i18n/fr] minor change in etypenavigation component

d-t-w

merge stable heads

[test, html validation] make validator selection somewhat smarter (at least it works properly when content is demoted from xhtml to html, making the XMLDemotingValidator class useless

minor code cleanup

merge

Backed out changeset 84b67ae41a0d: with this fix, some extra ]> appears in the browser. We need to fix the validator instead.

fix unittest broken by TrInfo schema change

check the validity of kwargs passed to match_rtype

merge stable heads

fix ticket #1382716 (problem was actually more subtle than I originally thought) * log errors happening in looping tasks and repo threads * do not swallow RegistryNotFound exceptions which occur during hook execution (only those happening during hook registry selection)

[web/views/basetemplates] nbsp breaks tests, use #160 instead

[web/views/workflow] hide tr_count attribute in wf transition form

backport stable

[html generation] even when doctype without extension are asked, we should define the xmlns:cubicweb attribute to avoid html validation error during tests (break tests with new calendar view in default branch since it demote to html)

handle the fact that there might be no views registered in some configurations (#1382716)

[ms] fix ms planner crash on some ambiguous queries, as demonstrated in a comment on http://www.cubicweb.org/ticket/1382452

[ms] #1382452: incorrect results with multi-sources when some relation is crossed among sources with non-invariant variables, we've to introduce an intermediary steps to fetch all relations from all sources, then make the join with intermediary tables containing possible variables values.