Re: "Final" version of tty child frames

[Gerd Möllmann]

Gerd Möllmann <gerd.moellmann@gmail.com> writes:

> Gerd Möllmann <gerd.moellmann@gmail.com> writes:
>
>> martin rudalics <rudalics@gmx.at> writes:
>>
>>> and emacs -Q -nw do C-l followed by M-l and C-l.  That is, make a child
>>> frame, switch to it and make it invisible.  This leaves the point (bar)
>>> cursor in the middle of the normal, root frame in an area where there's
>>> no text and you cannot really move it anywhere.  'xterm-mouse-mode' can
>>> help me out but this might not be available everywhere.
>>>
>>> Any ideas?
>>
>> The bar cursor must have been put where it is by redisplay. If the root
>> has no text where the bar cursor is, it must have been the display of
>> the child. Maybe the bar is still in the invisible child?
>>
>> That would fit with the fact that you can move the bar to the root frame
>> with the mouse IIUC, which should mean that C-x 5 o probably also works.
>>
>> I suspect make-frame-invisible has to do something. Do we have to
>> change selected-frame maybe? Or something like that?
>>
>> Please tell if you'd like me to take a look at this. I'm away for the
>> rest of the evening, so it would have to wait 'til tomorrow, alas.
>
> I've pulled now, and I can reproduce it here. Some observations.
>
> After C-l M-l the child frame is no longer visible, but the bar cursor
> is.
>
>   M-: (frame-visible-p tty-child-frame)
>   => t.
>
> That looks to me like something in the visibility department did not
> work, or worked only temporarily and was reverted (?). Which can also be
> seen with M-x redraw-display.
>
> ATM, I don't see where that could happen. Please let me know when you
> find something out.

I've used this to trace SET_FRAME_VISIBLE.

1 file changed, 5 insertions(+), 1 deletion(-)
src/frame.h | 6 +++++-

modified   src/frame.h
@@ -1433,8 +1433,10 @@ #define AUTO_FRAME_ARG(name, parameter, value)           
\
    if some changes were applied to it while it wasn't visible (and hence
    wasn't redisplayed).  */
 INLINE void
-SET_FRAME_VISIBLE (struct frame *f, bool v)
+xSET_FRAME_VISIBLE (struct frame *f, bool v, const char *file,
+                   int line, const char *func)
 {
+  fprintf (stderr, "%s:%d %s SET_FRAME_VISIBLE %p %d\n", file, line, func, f, 
v);
   if (v)
     {
       if (v == 1 && f->visible != 1)
@@ -1445,6 +1447,8 @@ SET_FRAME_VISIBLE (struct frame *f, bool v)
   f->visible = v;
 }

+#define SET_FRAME_VISIBLE(f, v) xSET_FRAME_VISIBLE(f, v, __FILE__, __LINE__, 
__func__)
+
 /* Set iconified status of frame F.  */
 INLINE void
 SET_FRAME_ICONIFIED (struct frame *f, int i)


Now C-l M-l C-l M-x C-g. I see a trace like

Sun Jan 12 07:35:45 CET 2025
frame.c:1246 make_initial_frame SET_FRAME_VISIBLE 0x155015c00 1
frame.c:1369 make_terminal_frame SET_FRAME_VISIBLE 0x15506ef00 1
frame.c:1781 do_switch_frame SET_FRAME_VISIBLE 0x155015c00 1
frame.c:1782 do_switch_frame SET_FRAME_VISIBLE 0x155015c00 1
frame.c:1781 do_switch_frame SET_FRAME_VISIBLE 0x155015c00 1
frame.c:1782 do_switch_frame SET_FRAME_VISIBLE 0x15506ef00 1
frame.c:1781 do_switch_frame SET_FRAME_VISIBLE 0x155015c00 1
frame.c:1782 do_switch_frame SET_FRAME_VISIBLE 0x155015c00 1
frame.c:1781 do_switch_frame SET_FRAME_VISIBLE 0x155015c00 1
frame.c:1782 do_switch_frame SET_FRAME_VISIBLE 0x15506ef00 1
frame.c:1781 do_switch_frame SET_FRAME_VISIBLE 0x155015c00 1
frame.c:1782 do_switch_frame SET_FRAME_VISIBLE 0x155015c00 1
frame.c:1781 do_switch_frame SET_FRAME_VISIBLE 0x155015c00 1
frame.c:1782 do_switch_frame SET_FRAME_VISIBLE 0x15506ef00 1
frame.c:3136 Fmake_frame_visible SET_FRAME_VISIBLE 0x15506ef00 1
frame.c:3194 Fmake_frame_invisible SET_FRAME_VISIBLE 0x15506ef00 0
^^^^^^^
This is from the last C-l, and below is M-x C-g
frame.c:3136 Fmake_frame_visible SET_FRAME_VISIBLE 0x155015c00 1
frame.c:1781 do_switch_frame SET_FRAME_VISIBLE 0x155015c00 1
...

So the child is invisible after the last C-l. Whether or not it becoming
visible again is something else to consider later, I think.