fix bug with algorithm and his hypothesis in case of new created study (not imported from python script)

This commit is contained in:
ptv 2009-01-13 13:04:50 +00:00
parent 95e7e98096
commit 9fa65460e9

View File

@ -1368,6 +1368,8 @@ bool _pyHypothesis::Addition2Creation( const Handle(_pyCommand)& theCmd,
algo = theGen->FindAlgo( myGeom, theMesh, this );
if ( algo.IsNull() )
return false;
// attach hypothesis creation command to be after algo creation command
// because it can be new created instance of algorithm
algo->GetCreationCmd()->AddDependantCmd( theCmd );
}
myIsWrapped = true;
@ -1386,10 +1388,10 @@ bool _pyHypothesis::Addition2Creation( const Handle(_pyCommand)& theCmd,
}
// set a new creation command
GetCreationCmd()->Clear();
// set dependance between creation and addition to mesh
// SetCreationCmd( theCmd );
GetCreationCmd()->AddDependantCmd( theCmd );
// replace creation command by wrapped instance
// please note, that hypothesis attaches to algo creation command (see upper)
SetCreationCmd( theCmd );
// clear commands setting arg values
list < Handle(_pyCommand) >::iterator argCmd = myArgCommands.begin();