rahuel [Wed, 9 Mar 2005 08:54:08 +0000 (08:54 +0000)]
IsValid and Valid methods have now an optionnal parameter kLoopSwitch which is true by default :
In some cases we do not need to check the validity of Loops and Switches
Memory Leaks
Check of Switches was added in Valid method
A graph must be loaded even if there is a uncorrect link (not created)
rahuel [Wed, 9 Mar 2005 08:50:01 +0000 (08:50 +0000)]
IsValid and Valid methods have now an optionnal parameter kLoopSwitch which is true by default :
In some cases we do not need to check the validity of Loops and Switches
asv [Mon, 7 Feb 2005 14:30:36 +0000 (14:30 +0000)]
Fix for PAL8028: calling to Editors data model only after "OK" button was pressed in EditPorts, EditFunction, AddPort operations. Pressing Cancel does not lead to switching data models.
asv [Mon, 7 Feb 2005 11:38:07 +0000 (11:38 +0000)]
Fix for PAL8027. from bugzilla:
The following program behaviour was implemented for "Add node", "Insert file", "Paste node" operations: if the dataflow is being executed - user is asked to Kill the execution before operation or cancel the operation.
changes in the code: SUPERVGUI_Main::ReadyToModify() operation was introduced to handle checking and killing of execution.
asv [Mon, 7 Feb 2005 06:27:21 +0000 (06:27 +0000)]
Fixed for bug PAL7960: calling Editing() in Main was moved "closer" to certain actions (in AddNode, only when "Add node" button is pressed but not before dialog was opened, etc.).
asv [Thu, 3 Feb 2005 12:52:33 +0000 (12:52 +0000)]
Improvement for handling node events coming from Engine. SUPERVGUI_Thread::run(), Main::execute(), etc. methods were improved, several unsused methods were removed.
asv [Wed, 2 Feb 2005 13:33:29 +0000 (13:33 +0000)]
Fix for 7950. myExecuted boolean field added to the class. set to true on first execution. It is used in putDataStudy() -- if the dataflow was never executed, then the corresponding "Run xxx" SObject is not added to the Study.
asv [Mon, 31 Jan 2005 08:02:23 +0000 (08:02 +0000)]
Fix for bug PAL7854: Editing() method is called on Main and Graph_Impl not BEFORE run() and stepBystep() commands, but inside of them. And if the dataflow is being executed - then it is NOT called.
asv [Mon, 31 Jan 2005 06:00:04 +0000 (06:00 +0000)]
Fix for bug PAL78476: when out-port is added to study, it must try to put its value to study right at the moment of "put" command. After that, it will be added to study in Main::execute() method on "Done" event from a node.
asv [Fri, 28 Jan 2005 12:46:13 +0000 (12:46 +0000)]
Fix for bug PAL7845: main->Editing() is called NOT before opening "Browse" node dialog or "Set Value" of input port dialog, but only after pressing "OK" in these dialogs.
asv [Thu, 27 Jan 2005 08:34:19 +0000 (08:34 +0000)]
fix for 7817: in removeArrayChild() the returned list of children of myArray is checked for NULL. when no children exist, then it will be NULL and it was not checked.
asv [Thu, 27 Jan 2005 08:17:22 +0000 (08:17 +0000)]
Fix for PAL7823 (checking for if study is locked before doing New Dataflow command).
Also some re-engineering of the code: new function createDataflow() is introduced, it combines functionality of the 3 commands: newDataflow(), importDataflow(), modifyDataflow(). These 3 functions did basically the same thing, with slight differences, which are implemented in createDataflow() using 'mode' parameter.
asv [Thu, 27 Jan 2005 06:13:30 +0000 (06:13 +0000)]
The graph is modified in order to be executable. Invalid link from the first node (some obj_ref out-port) to a Gate port of the second node is removed.
asv [Wed, 26 Jan 2005 13:38:37 +0000 (13:38 +0000)]
Fix for PAL7164: after analysis of the bug, it was found out that PublishInStudy() in GEOM was called twice from Supervisor. It was done from CanvasPort::sync() function. Since sync() is an asinchronious function, it is called when a signal comes, and in other occasions, it was decided to move the put_in_study functionality to another place - to execute() function of Main, which handles events coming from Engine.
If an event is "Node has successfully finished execution", then its CanvasOutPorts are analized, and if some port has "myIsInStudy" flag - then value of this port is put into study.
It fixes the bug, since execute() is called exactly once on every event (controlled by Supervisor Engine).
asv [Wed, 26 Jan 2005 13:33:35 +0000 (13:33 +0000)]
Changes in getPopupMenu() memthod:
myDeleteItem was removed from the class, it was deactivated in CanvasCellNode - successor of CanvasNode, now it is not even added to CanvasNode's menu if the view_type is TABLE.
asv [Wed, 26 Jan 2005 13:31:48 +0000 (13:31 +0000)]
Commented out setNodeName() methods of CanvasCellNode and CanvasCellEndNodes, and getPopupMenu() method of CanvasCellNode.
setNodeName() : decided to remove all node "edition" operations from CellNodes (delete node, rename node), becase it produces bugs in unsynchronization between CanvasNode-s and CanvasCellNode-s.
getPopupMenu() : before myDeleteItem was deactivated in CanvasCellNode, now it is not even added to CanvasNode's menu if the view_type is TABLE.
asv [Mon, 24 Jan 2005 13:12:34 +0000 (13:12 +0000)]
FindContainer() method does not exist any more. FindOrLoad_Component() starts the container for us. If it fails to start the container, then it returns nil component. We don't need to check for if the container was successfully started.
