* unix/tkUnixPort.h: Disabled inclusion of the private Tcl header

        * win/tkWinPort.h:      file tclInt.h.  Tk ought to have a tiny and
        shrinking number of calls of private Tcl routines.  Each Tk source
        file doing this should follow the convention in the macosx port and
        have its own #include "tclInt.h".

        * generic/tkEvent.c:    Disabled calls to private Tcl routine
        TclInExit().  See comment in TkCreateExitHandler() for full rationale.

1 files changed, 23 insertions, 3 deletions

diff --git a/generic/tkEvent.c b/generic/tkEvent.c
index fa81b65..ceffd0a 100644
--- a/generic/tkEvent.c
+++ b/generic/tkEvent.c
@@ -12,7 +12,7 @@
  * See the file "license.terms" for information on usage and redistribution of
  * this file, and for a DISCLAIMER OF ALL WARRANTIES.
  *
- * RCS: @(#) $Id: tkEvent.c,v 1.29 2005/09/21 10:54:40 dkf Exp $
+ * RCS: @(#) $Id: tkEvent.c,v 1.30 2005/09/28 18:31:57 dgp Exp $
  */
 
 #include "tkPort.h"
@@ -1880,7 +1880,26 @@ TkCreateExitHandler(
     exitPtr->proc = proc;
     exitPtr->clientData = clientData;
     Tcl_MutexLock(&exitMutex);
-    if (firstExitPtr == NULL && !TclInExit()) {
+
+    /*
+     * The call to TclInExit() is disabled here.  That's a private
+     * Tcl routine, and calling it is causing some trouble with portability
+     * of building Tk.  We should avoid private Tcl routines generally.
+     *
+     * In this case, the TclInExit() call is being used only to prevent
+     * a Tcl_CreateExitHandler() call when Tcl finalization is in progress.
+     * That's a situation that shouldn't happen anyway.  Recent changes
+     * within Tcl_Finalize now cause a Tcl_Panic() to happen if exit
+     * handlers get added after exit handling is complete.  By disabling
+     * the guard here, that panic will serve to help us find the buggy
+     * conditions and correct them.
+     *
+     * We can restore this guard if we find we must (hopefully getting
+     * public access to TclInExit() if we discover extensions really do
+     * need this), but during alpha development, this is a good time to
+     * dig in and find the root causes of finalization bugs.
+     */
+    if (firstExitPtr == NULL/* && !TclInExit()*/) {
 	Tcl_CreateExitHandler(TkFinalize, NULL);
     }
     exitPtr->nextPtr = firstExitPtr;
@@ -1958,7 +1977,8 @@ TkCreateThreadExitHandler(
     exitPtr = (ExitHandler *) ckalloc(sizeof(ExitHandler));
     exitPtr->proc = proc;
     exitPtr->clientData = clientData;
-    if (tsdPtr->firstExitPtr == NULL && !TclInExit()) {
+    /* See comments in TkCreateExitHandler */
+    if (tsdPtr->firstExitPtr == NULL/* && !TclInExit()*/) {
 	Tcl_CreateThreadExitHandler(TkFinalizeThread, NULL);
     }
     exitPtr->nextPtr = tsdPtr->firstExitPtr;