Changeset 8e40da74

Timestamp:

Sep 27, 2011, 4:16:21 PM (13 years ago)

Author:

David Benjamin <davidben@mit.edu>

Children:

bc6d81d

Parents:

bbd0cf1

git-author:

David Benjamin <davidben@mit.edu> (09/27/11 16:08:44)

git-committer:

David Benjamin <davidben@mit.edu> (09/27/11 16:16:21)

Message:

Fix dirtying windows inside a redraw handler

This is a seriously sketchy thing to do, but at least maintain
dirty and dirty_subtree bits correctly.

Tested by triggering an editwin dirty from the sepbar code.

File:

• window.c (modified) (1 diff)

window.c

@@ -441 +441 @@
 }
 
+static bool _owl_window_is_subtree_dirty(owl_window *w)
+{
+  owl_window *child;
+
+  if (w->dirty)
+    return true;
+  for (child = w->child;
+       child != NULL;
+       child = child->next) {
+    if (child->dirty_subtree)
+      return true;
+  }
+  return false;
+}
+
 static void _owl_window_redraw_subtree(owl_window *w)
 {
   FuncOneArg ptr = (FuncOneArg)_owl_window_redraw_subtree;
+
   if (!w->dirty_subtree)
     return;
+
   _owl_window_redraw(w);
   owl_window_children_foreach(w, first_arg_only, &ptr);
-  w->dirty_subtree = 0;
+
+  /* Clear the dirty_subtree bit, unless a child doesn't have it
+   * cleared because we dirtied a window in redraw. Dirtying a
+   * non-descendant window during a redraw handler is
+   * discouraged. Redraw will not break, but it is undefined whether
+   * the dirty is delayed to the next event loop iteration. */
+  if (_owl_window_is_subtree_dirty(w)) {
+    owl_function_debugmsg("subtree still dirty after one iteration!");
+  } else {
+    w->dirty_subtree = 0;
+  }
 }