Hi, the latest version of 21.2 from CVS generates huge numbers of warnings under egcs-1.0.3, because the initialiser for the lrecord_header member of Lisp_Subr in lisp.h and of I_hate_C in symeval.h is wrong. Patch and ChangeLog attached. Robert 1999-05-04 Robert Pluim <rpluim(a)bigfoot.com&gt; * symeval.h (symbol_value_forward_lheader_initializer): remove extra initializer * lisp.h (subr_lheader_initializer): remove extra initializer Index: lisp.h =================================================================== RCS file: /usr/CVSroot/XEmacs/xemacs/src/lisp.h,v retrieving revision 1.38.2.9 diff -u -w -r1.38.2.9 lisp.h --- lisp.h 1999/04/22 11:10:16 1.38.2.9 +++ lisp.h 1999/05/04 23:07:11 ＠＠ -1545,7 +1545,7 ＠＠ /* Can't be const, because then subr->doc is read-only and Snarf_documentation chokes */ -#define subr_lheader_initializer { 0, 0, 0 } +#define subr_lheader_initializer { 0, {0} } #define DEFUN(lname, Fname, min_args, max_args, prompt, arglist) \ Lisp_Object Fname (EXFUN_##max_args); \ Index: symeval.h =================================================================== RCS file: /usr/CVSroot/XEmacs/xemacs/src/symeval.h,v retrieving revision 1.4.2.2 diff -u -w -r1.4.2.2 symeval.h --- symeval.h 1999/04/22 07:27:28 1.4.2.2 +++ symeval.h 1999/05/04 23:07:12 ＠＠ -293,7 +293,7 ＠＠ void defvar_magic (CONST char *symbol_name, CONST struct symbol_value_forward *magic); -#define symbol_value_forward_lheader_initializer { 1, 0, 0 } +#define symbol_value_forward_lheader_initializer { 1, {0} } #define DEFVAR_SYMVAL_FWD(lname, c_location, forward_type, magicfun) do { \ static CONST_IF_NOT_DEBUG struct symbol_value_forward I_hate_C \