Re: buffer.c/buffer.h: How to add new buffer-local variables?

[Keith David Bershatsky]

Is it possible that a038df77de7b1aa2d73a6478493b8838b59e4982 broke the ability 
for Emacs to maintain the value of certain WINDOW Lisp_Object defined in 
window.h?

Here is a snippet from a modified window.h.  All of the WINDOW caches for the 
fake cursors are coming back as INVALID_LISP_OBJECT.  A test for NILP on those 
invalid lisp objects comes back TRUE, and subsequent functions fail when trying 
to process the object that reports as being "DEAD" in gdb printouts.  This 
happens on the NS and w32 ports of Emacs.  I have not tested X11, but I assume 
the result will be the same.

If it is highly unlikely that a038df77de7b1aa2d73a6478493b8838b59e4982 is the 
culprit, then I would imagine that the culprit is somewhere between March 28, 
2019 and today.  If anyone would like me to start going back in time to test 
other comments between March 28, 2019 and today, please let me know ....

    /* An alist with parameters.  */
    Lisp_Object window_parameters;

    /* The help echo text for this window.  Qnil if there's none.  */
    Lisp_Object mode_line_help_echo;


/* *************************************************************************** 
*/
/* MULTIPLE-CURSORS */

    Lisp_Object mc_temp_cache;

    /* The cache for multiple fake cursors. */
    Lisp_Object mc_cache;

    /* The cache for crosshairs. */
    Lisp_Object ch_cache;

    /* The cache for visible fill column. */
    Lisp_Object fc_cache;

    /* L.S.L. color vector used by crosshairs. */
    Lisp_Object ch_foreground;

    /* L.S.L. color vector used by visible fill column. */
    Lisp_Object fc_foreground;

/* *************************************************************************** 
*/


    /* No Lisp data may follow this point; mode_line_help_echo must be
       the last Lisp member.  */

    /* Glyph matrices.  */
    struct glyph_matrix *current_matrix;
    struct glyph_matrix *desired_matrix;