Created by Jukka Santala (donwulff@iki.fi) on Tuesday, April 7, 1998 4:56:14 AM PDT Additional Details : The call stack backtrace tells more than a thousand words : (It''s an assert instead of prot. fault because this has the below patch compiled in; it doesn''t affect the actual bug, just lets me recover from the error gracefully) XP_AssertAtLine(char * 0x00866778, int 63) line 2649 CFormButton::SetContext(CAbstractCX * 0x00000000) line 63 + 36 bytes CFormElement::GetFormElement(CAbstractCX * 0x00000000, LO_FormElementData_struct * 0x00a5c8c0) line 135 FE_FreeFormElement(MWContext_ * 0x009ebd30, LO_FormElementData_struct * 0x00a5c8c0) line 720 + 11 bytes lo_FreeDocumentFormListData(MWContext_ * 0x009ebd30, lo_SavedFormListData_struct * 0x00a2d5e0) line 858 + 19 bytes LO_FreeDocumentFormListData(MWContext_ * 0x009ebd30, void * 0x00a2d5e0) line 6387 + 13 bytes SHIST_FreeHistoryEntry(MWContext_ * 0x009ebd30, _History_entry * 0x00a29c30) line 101 + 16 bytes SHIST_EndSession(MWContext_ * 0x009ebd30) line 438 + 13 bytes CAbstractCX::DestroyContext() line 116 + 740 bytes CDCCX::DestroyContext() line 186 CPaneCX::DestroyContext() line 304 CWinCX::DestroyContext() line 619 CGenericView::FrameClosing() line 215 CMainFrame::OnClose() line 660 CWnd::OnWndMsg(unsigned int 16, unsigned int 0, long 0, long * 0x0012fd3c) line 1815 CWnd::WindowProc(unsigned int 16, unsigned int 0, long 0) line 1575 + 30 bytes AfxCallWndProc(CWnd * 0x00399ff0 {CMainFrame hWnd=0x0009035a}, HWND__ * 0x0009035a, unsigned int 16, unsigned int 0, long 0) line 217 + 26 bytes AfxWndProc(HWND__ * 0x0009035a, unsigned int 16, unsigned int 0, long 0) line 371 AfxWndProcBase(HWND__ * 0x0009035a, unsigned int 16, unsigned int 0, long 0) line 203 + 21 bytes USER32! 77e72cc4() USER32! 77e72ce2() CNetscapeStatusBar::CParentSubclass::ParentSubclassProc(HWND __ * 0x0009035a, unsigned int 16, unsigned int 0, long 0) line 1077 + 29 bytes USER32! 77e713ed() Temporary patch (Munched by the form): diff --unified --recursive mozorg/ns/cmd/winfe/fmbutton.cpp mozilla/ns/cmd/winfe/fmbutton.cpp --- mozorg/ns/cmd/winfe/fmbutton.cpp Mon Mar 30 17:57:41 1998 +++ mozilla/ns/cmd/winfe/fmbutton.cpp Mon Apr 06 18:42:09 1998 @@ -60,8 +60,9 @@ // Call the base. CFormElement::SetContext(pCX); + XP_ASSERT(pCX); //Update the widget if present for correct callbacks. - if(m_pWidget) { + if(m_pWidget&&pCX) { m_pWidget->RegisterContext(pCX->GetContext()); } } This will solve the immediate problem, but as to how and why pCX gets to be null is another story.

It gets NULL I'm guessing because earlier in the stack it is being deleted. Handing over to forms expert to evaluate the patch. I think it's good.

The context is zero because someone higher up the call stack didn't want to have to be bothered by no stinking contexts, so it sent in a zero. donwulff's patch is fine under the circumstances, but I'm going to skip it because law@netscape.com already patched it one call higher up -- in GetFormElement, on 16 July 98. I say "already": Jukka saw this problem three months earlier. Sorry about the delay. I plead distraction; we plan to be more responsive in the future.

fixed on old codebase -- marking verified

width of field QU_CONTENT_OID is too small in table SD_QUOTE_DETAIL