In the past week, a number of minor bugs have been fixed: d1a65dc0 Check in install-sh as synced from automake (#110368) 20a53df1 autoreconf: Support AM_GNU_GETTEXT_REQUIRE_VERSION a8b69e68 Fix more bugs in specific tests under AC_LANG(C++). 8e540139 Make ‘forbidden tokens, basic’ test more robust. 0f7e5a54 Do not apply --program-transform-name to build-aux scripts. c8b3f698 AC_FUNC_STRERROR_R: Include string.h in test program. aa27fc57 Define AC_REQUIRE_AUX_FILE with AC_DEFUN. 15ada9d3 _AC_PROG_YYTEXT_DECL: Forward declare yywrap (#110312) I wish to thank Jannick for debugging assistance, Michał Górny for submitting a patch five years ago that was only just now applied, and Nicholas Guriev for bringing that patch to my attention. Commit 3b411849 is not exactly a bug fix but if you're reading this mailing list you probably want to know about it. As of the 2020-11-07 update, config.sub and config.guess unconditionally use $(...) command substitution; see <https://lists.gnu.org/archive/html/config-patches/2020-11/msg00011.html>. Therefore I have added this to the set of required shell features that _AS_DETECT_BETTER_SHELL looks for. This will ensure that when AC_CANONICAL_* invoke these scripts, they use a shell that can handle $(...). $(...) is approximately the same vintage as shell functions (both are present in SUSv2, see https://pubs.opengroup.org/onlinepubs/7990989775/xcu/chap2.html#tag_001_006_003); there *are* shells that implement one but not the other, but I don't think we need to worry about OSes that don't have *any* shell that implements both. We still have seven release blockers and I still need help with them. This list is in descending order of priority, and, sadly, also in descending order of difficulty. #110350 AX_PROG_CC_FOR_BUILD broken with 2.69c #110294 AC_LANG_PUSH/AC_LANG_POP malfunction inside AC_DEFUN #110349 AC_INIT quoting headaches #110305 Replace autom4te output file atomically #110312 2.69b: broken AC_PROG_LEX macro (unreproducible) #110359 PATH ignored searching for g++ (unreproducible) #110347 Revise documentation of when configure enters cross-compilation mode. I will be happy to consider a patch for any other open bug as well, but depending on how invasive it is, it might have to wait for 2.71. Finally, please do continue to test 2.69d and report more bugs. zw
