Wed, 14 Apr 2010 17:38:24 +0200 [web session] fix web session id bug on automatic reconnection. The web session id should keep the first connection id, then differ of the repo connection id once some reconnection has been done (since the session cookie isn't updated in such cases). Also, use a lock to avoid potential race condition on reconnection.
Sylvain Thénault <sylvain.thenault@logilab.fr> [Wed, 14 Apr 2010 17:38:24 +0200] rev 5251
[web session] fix web session id bug on automatic reconnection. The web session id should keep the first connection id, then differ of the repo connection id once some reconnection has been done (since the session cookie isn't updated in such cases). Also, use a lock to avoid potential race condition on reconnection.
Wed, 14 Apr 2010 17:31:41 +0200 [packaging] 3.8 depends on lgc 0.50 (new argument to dot generator in lgc.graph)
Sylvain Thénault <sylvain.thenault@logilab.fr> [Wed, 14 Apr 2010 17:31:41 +0200] rev 5250
[packaging] 3.8 depends on lgc 0.50 (new argument to dot generator in lgc.graph)
Wed, 14 Apr 2010 16:26:34 +0200 backport stable
Sylvain Thénault <sylvain.thenault@logilab.fr> [Wed, 14 Apr 2010 16:26:34 +0200] rev 5249
backport stable
Wed, 14 Apr 2010 16:26:20 +0200 [views] fix bug on gc debugging view stable
Sylvain Thénault <sylvain.thenault@logilab.fr> [Wed, 14 Apr 2010 16:26:20 +0200] rev 5248
[views] fix bug on gc debugging view
Wed, 14 Apr 2010 16:26:02 +0200 [testlib] ctrl_publish now allow to specify the controller id as argument stable
Sylvain Thénault <sylvain.thenault@logilab.fr> [Wed, 14 Apr 2010 16:26:02 +0200] rev 5247
[testlib] ctrl_publish now allow to specify the controller id as argument
Wed, 14 Apr 2010 11:26:36 +0200 [web] stop having actions configurable through cwproperties: this clutter site management and .po files for a useless fonctionnality (which, where and how actions are displayed is handled by code)
Sylvain Thénault <sylvain.thenault@logilab.fr> [Wed, 14 Apr 2010 11:26:36 +0200] rev 5246
[web] stop having actions configurable through cwproperties: this clutter site management and .po files for a useless fonctionnality (which, where and how actions are displayed is handled by code)
Wed, 14 Apr 2010 10:31:09 +0200 [testlib] we should not replace db-name in sources, this cause undesired side effect (when running pytest on whole cw for instance). Proper fix in logilab.database stable
Sylvain Thénault <sylvain.thenault@logilab.fr> [Wed, 14 Apr 2010 10:31:09 +0200] rev 5245
[testlib] we should not replace db-name in sources, this cause undesired side effect (when running pytest on whole cw for instance). Proper fix in logilab.database
Wed, 14 Apr 2010 10:29:38 +0200 [web] put a fake object that raise Unauthorized on any attribute access as req.cnx and req._user, so we are properly asked to authenticated on any view that tries to do something with one of those attributes (instead of doing defensive programming everywhere we're doing that)
Sylvain Thénault <sylvain.thenault@logilab.fr> [Wed, 14 Apr 2010 10:29:38 +0200] rev 5244
[web] put a fake object that raise Unauthorized on any attribute access as req.cnx and req._user, so we are properly asked to authenticated on any view that tries to do something with one of those attributes (instead of doing defensive programming everywhere we're doing that)
Wed, 14 Apr 2010 10:25:51 +0200 [etwist] we must call all error not handled in _render_request, else the thread fail silently and nothing ends up on the ui
Sylvain Thénault <sylvain.thenault@logilab.fr> [Wed, 14 Apr 2010 10:25:51 +0200] rev 5243
[etwist] we must call all error not handled in _render_request, else the thread fail silently and nothing ends up on the ui
Wed, 14 Apr 2010 09:38:08 +0200 backport stable
Sylvain Thénault <sylvain.thenault@logilab.fr> [Wed, 14 Apr 2010 09:38:08 +0200] rev 5242
backport stable
(0) -3000 -1000 -300 -100 -10 +10 +100 +300 +1000 +3000 tip