APPROVE COMMIT 21.5
I'm suspicious of the computation of most-postive-fixnum (see last
hunk). I don't think it's relevant on any real architecture we deal
with, though.
# HG changeset patch
# Parent d58b1c68ec10a5f5c29361ec82f9376a4073662d
Fix typos in comments.
diff -r d58b1c68ec10 src/alloc.c
--- a/src/alloc.c Tue Nov 14 14:14:25 2017 +0900
+++ b/src/alloc.c Tue Nov 14 14:25:45 2017 +0900
@@ -3499,7 +3499,7 @@
/* Build a string whose content is a translatable message, and translate
the message according to the current language environment.
String must be pure-ASCII, and when compiled with error-checking,
- an abort will have if not pure-ASCII. */
+ an abort will happen if not pure-ASCII. */
Lisp_Object
build_msg_ascstring (const Ascbyte *str)
diff -r d58b1c68ec10 src/lisp.h
--- a/src/lisp.h Tue Nov 14 14:14:25 2017 +0900
+++ b/src/lisp.h Tue Nov 14 14:25:45 2017 +0900
@@ -1040,7 +1040,7 @@
DECLARE_BAD_INTCLASS_ARITH_COMPARE (bytecl, Membpos) \
DECLARE_BAD_INTCLASS_ARITH_COMPARE (bytecl, Memxpos)
-/* Declare the "bad" or disallowed arithmetic and comparion operations
+/* Declare the "bad" or disallowed arithmetic and comparison operations
between class BYTECL (a mem class) and various non-mem classes.
Meant to go inside the private section of class GOOD. */
@@ -1798,6 +1798,7 @@
#define FIXNUM_VALBITS (BITS_PER_EMACS_INT - FIXNUM_GCBITS)
#define VALBITS (BITS_PER_EMACS_INT - GCBITS)
+/* #### Does this actually work? EMACS_INT might be "long long"! */
#define MOST_POSITIVE_FIXNUM_UNSIGNED ((1UL << (FIXNUM_VALBITS - 1)) -1UL)
#define MOST_POSITIVE_FIXNUM ((EMACS_INT) MOST_POSITIVE_FIXNUM_UNSIGNED)
#define MOST_NEGATIVE_FIXNUM (-(MOST_POSITIVE_FIXNUM) - 1)