Address the easy test failures in tests/automated. src/ChangeLog addition: 2011-03-11 Aidan Kehoe <kehoea@parhasard.net> * bytecode.c (optimize_byte_code): Only transform assignments to keywords to Bdiscard if NEED_TO_HANDLE_21_4_CODE is turned on. Cf. similar code in reject_constant_symbols(). tests/ChangeLog addition: 2011-03-11 Aidan Kehoe <kehoea@parhasard.net> * automated/byte-compiler-tests.el: (defconst :foo 1) now gives a warning when byte-compiled, check for that. (setq :foo 1) now errors with interpreted code, but succeeds with byte-compiled code; check for the former, wrap a Known-Bug-Expect-Failure around a check for the error in the latter case, we can't yet remove this behaviour while we're using packages compiled by 21.4. * automated/lisp-tests.el (wrong-type-argument): Integer zero is a valid argument to #'substring-no-properties, use Assert not Check-Error for it. Check some other aspects of the functionality of #'substring-no-properties in passing.

/* XPM */
static char *face[] = {
/* width height num_colors chars_per_pixel */
"    17    17        4            1",
/* colors */
". c #000000",
"a c #a8b038",
"b c #f8f800",
"c s None c None",
/* pixels */
"ccccccccccccccccc",
"ccccccccccccccccc",
"cccccbbbbbbcccccc",
"ccccbbbbbbbbccccc",
"cccbbbbbbbbbbcccc",
"ccbbbbbbbbbbbbccc",
"cbbaaabbbbaaabbcc",
"cbbbbbbbbbbbbbbcc",
"cbbbbbbaabbbbbbcc",
"cbbbbbbaabbbbbbcc",
"cbbbbbbaabbbbbbcc",
"cbbbbbaaaabbbbbcc",
"ccbbabbbbbbabbccc",
"cccbbabbbbabbcccc",
"ccccbbaaaabbccccc",
"cccccbbbbbbcccccc",
"ccccccccccccccccc"
};