blob: 1048bd96ef753996203837998cd1481e522ae8ca [file] [log] [blame]
cws 'atkbridge'
#Issue number: i#47890#
Submitted by: mmeeks
Hacked up prototype of atk bridge
Serious problems
+ Threading/locking:
+ incoming CORBA calls & the GDK lock
+ how are these being processed & on what thread ?
+ are we holding the GDK_THREADS lock ?
+ can we even do that ?
+ is it really necessary to be thread safe ?
+ how does this work in combination with the (unsafe) GAIL code ?
+ what should incoming CORBA calls be doing ?
+ esp. since we can't tell if they're coming from
in-proc or not either [ though this is unlikely ]
Test:
+ in-line text editing, does the TEXT_CHANGED signal get it right,
+ why not copy/paste/delete etc. ?
+ check vs. writer & other bits ...
+ AtkSelection
+ AtkHyper*
* At-poke
+ implement non-gui mode - for to-console event logging
+ logging
+ more detail from remaining events
+ add a Tree navigation thing instead (?)
+ poke a sub-child (?)
+ embed a tree widget inside the tree view ?
+ AtkHyperText testing (?)
Known bugs:
+ AtkText
+ selection interface - multiple selections ?
+ word boundary issues
+ copy AccessibleTextImpl.java's getAfterIndex eg.
+ the 'getFoo' methods need to use UNO_QUERY_THROW &
throw an exception to avoid null pointer dereferences.
+ AtkAttributeSet (etc.)
+ AtkEditableText
+ finish/test AtkTable
+ HyperLink 'link_activated', HyperText 'link_selected' (?)
+ tooltips create new toplevels with broken roles.