BUG: in compilation or 21.5(beta9)

As a result of edits to eval.c src/ChangeLog: 2009-08-11 Aidan Kehoe <kehoea(a)parhasard.net&gt; * eval.c (For): Ignore multiple values when comparing with Qnil, but pass any multiple values back for the last arg. * eval.c (Fand): Ditto. On my Mac G5 (OS X 10.5.8) I now see gcc-4.2 -c -Wall -Wno-switch -Wundef -Wsign-compare -Wno-char-subscripts -Wpacked -Wpointer-arith -Wunused-parameter -g -g -Os -pipe -mcpu=970 -mtune=970 -mpim-altivec -mpowerpc-gpopt -mpowerpc64 -Demacs -I. -I/Users/royar/src/xemacs-hg/src -DHAVE_CONFIG_H -I/usr/X11/include -I/usr/X11/include/freetype2 -I/opt/local/include -I/usr/X11/include eval.c eval.c: In function 'For': eval.c:845: error: address of register variable 'val' requested eval.c: In function 'Fand': eval.c:877: error: address of register variable 'val' requested make[1]: *** [eval.o] Error 1 make: *** [src] Error 2 The code reads REGISTER Lisp_Object val; LIST_LOOP_3 (arg, args, tail) { if (!NILP (IGNORE_MULTIPLE_VALUES (val = Feval (arg)))) I believe the problem is that Lisp_Object is a union in my compilation, not an int. The working version of XEmacs I am running is XEmacs 21.5 (beta29) "garbanzo" e95ddfd6a409 [Lucid] (powerpc-apple-darwin9.8.0, Mule) of Tue Aug 11 2009 on grendel.frinabulax.org The source for the above in that version is if (!NILP (val = Feval (arg))) return val; Removing "IGNORE_MULTIPLE_VALUES" from the two functions (For and Fand) allows the compilation to continue. I do not know what effects that removal has on the stability of the final program. -- Dr. Robert Delius Royar Associate Professor of English Morehead State University Morehead, Kentucky _______________________________________________ XEmacs-Beta mailing list XEmacs-Beta(a)xemacs.org http://calypso.tux.org/cgi-bin/mailman/listinfo/xemacs-beta