Not really a bug, but...

[Steven Swerling]

Hi Rob,

I've been plodding along w/WxMorphicPanel, things seem to be going well,
knock on wood.

I did get some crashes during image save since my subclass of
TopLevelPresenter does not yet know how to "saveView" or "restoreView".

Somehow, one of my test windows was in AllTopLevelPresenters even though
it was no longer alive. I managed to get quite a few crashes during the
early stages of my work w/ WxMorphicPanel. Somehow during this phase I
must have got a bug that didn't crash the image but managed to mangle
the 'deleteObject' machinery.

Anyway, the little snippet below fixed the problem. Just does a #isValid
check on the view's handle before calling storeView.

!TopLevelPresenter class methodsFor: 'as yet unclassified' stamp: 'sps
3/4/2005 13:22'!
shuttingDown: quitting

WxBase removeDeletedObjects.

self allTopLevelPresenters copy do: [:presenter |
presenter view handle isValid
ifTrue:[ presenter storeView ]
ifFalse: [self allTopLevelPresenters remove: presenter]
].
! !

[Rob Gayvert]

Post by Steven Swerling
Hi Rob,
I've been plodding along w/WxMorphicPanel, things seem to be going
well, knock on wood.
I did get some crashes during image save since my subclass of
TopLevelPresenter does not yet know how to "saveView" or "restoreView".
Somehow, one of my test windows was in AllTopLevelPresenters even
though it was no longer alive. I managed to get quite a few crashes
during the early stages of my work w/ WxMorphicPanel. Somehow during
this phase I must have got a bug that didn't crash the image but
managed to mangle the 'deleteObject' machinery.
Anyway, the little snippet below fixed the problem. Just does a
#isValid check on the view's handle before calling storeView.
!TopLevelPresenter class methodsFor: 'as yet unclassified' stamp: 'sps
3/4/2005 13:22'!
shuttingDown: quitting
WxBase removeDeletedObjects.
self allTopLevelPresenters copy do: [:presenter |
presenter view handle isValid
ifTrue:[ presenter storeView ]
ifFalse: [self allTopLevelPresenters remove: presenter]
].
! !

Good catch. I'm adding more null pointer checks in the primitives to
prevent the VM from crashing, but we'll probably still need isValid
checks at points like this to make it clean up properly.