Re: [Emacs-diffs] master 5c9304e: Disable some display optimizations when frames need redisplay

[Stefan Monnier]

> Tell you what: if you add commentary that describes the logic of using
> these two variables, the window's and frame's 'redisplay' flags, and
> how all those are used to decide which parts need to be redisplayed,
> and if that commentary makes it clear that the code I added is
> redundant, I will remove the code I added, and see if some alternative
> design would be better.  Deal?

How do you like the patch below?


        Stefan


diff --git a/src/xdisp.c b/src/xdisp.c
index 1524783..439d9cb 100644
--- a/src/xdisp.c
+++ b/src/xdisp.c
@@ -434,22 +434,47 @@ static Lisp_Object Vmessage_stack;
 
 static bool message_enable_multibyte;
 
-/* Nonzero if we should redraw the mode lines on the next redisplay.
-   If it has value REDISPLAY_SOME, then only redisplay the mode lines where
-   the `redisplay' bit has been set.  Otherwise, redisplay all mode lines
-   (the number used is then only used to track down the cause for this
-   full-redisplay).  */
+/* At each redisplay cycle, we should refresh everything there is to refresh.
+   To do that efficiently, we use many optimizations that try to make sure we
+   don't waste too much time updating things that haven't changed.
+   The coarsest such optimization is that, in the most common cases, we only
+   look at the selected-window.
+
+   To know whether other windows should be considered for redisplay, we use the
+   variable windows_or_buffers_changed: as long as it is 0, it means that we
+   have not noticed anything that should require updating anything else than
+   the selected-window.  If it is set to REDISPLAY_SOME, it means that since
+   last redisplay, some changes have been made which could impact other
+   windows.  To know which ones need redisplay, every buffer, window, and frame
+   has a `redisplay' bit, which (if true) means that this object needs to be
+   redisplayed.  If windows_or_buffers_changed is 0, we know there's no point
+   looking for those `redisplay' bits (actually, there might be some such bits
+   set, but then only on objects which aren't displayed anyway).
+
+   Mostly for historical reasons, windows_or_buffers_changed can also take
+   other non-zero values.  In that case, the precise value doesn't matter (it
+   encodes the cause of the setting but is only used for debugging purposes),
+   and what it means is that we shouldn't pay attention to any `redisplay' bits
+   and we should simply try and redisplay every since window out there.  */
 
-int update_mode_lines;
+int windows_or_buffers_changed;
 
-/* Nonzero if window sizes or contents other than selected-window have changed
-   since last redisplay that finished.
-   If it has value REDISPLAY_SOME, then only redisplay the windows where
-   the `redisplay' bit has been set.  Otherwise, redisplay all windows
-   (the number used is then only used to track down the cause for this
-   full-redisplay).  */
+/* Nonzero if we should redraw the mode lines on the next redisplay.
+   Similarly to `windows_or_buffers_changed', If it has value REDISPLAY_SOME,
+   then only redisplay the mode lines in those buffers/windows/frames where the
+   `redisplay' bit has been set.
+   For any other value, redisplay all mode lines (the number used is then only
+   used to track down the cause for this full-redisplay).
+
+   The `redisplay' bits are the same as those used for
+   windows_or_buffers_changed, and setting windows_or_buffers_changed also
+   causes recomputation of the mode lines of all those windows.  IOW this
+   variable only has an effect if windows_or_buffers_changed is zero, in which
+   case we should only need to redisplay the mode-line of those objects with
+   a `redisplay' bit set but not the window's text content (tho we may still
+   need to refresh the text content of the selected-window).  */
 
-int windows_or_buffers_changed;
+int update_mode_lines;
 
 /* True after display_mode_line if %l was used and it displayed a
    line number.  */