[Following up on <http://lists.gnu.org/archive/html/autoconf/2010-11/msg00036.html>, and adding bug-autoconf] On 11/23/2010 08:58 PM, Luke Mewburn wrote: > Is there any reason that AT_CHECK doesn't close this fd before > executing the application? Thanks for reporting the problem. I don't know of any reason AT_CHECK wouldn't do that, but I'm not intimately familiar with the code. Perhaps Paolo (who contributed this code back in 2003) can comment. > I've attached a patch that performs this, against autoconf-2.68. That patch doesn't smell right to me. The output is being redirected for a reason, and simply undoing the redirection feels odd. I suspect that any fix should be done inside AT_CHECK instead. Could you try the following instead, and see whether it works for you? I'm not sure it's right (hadn't had time to analyze this code) but I think it's an improvement. There's a similar issue with high FDs when 'configure' invokes compilers and whatnot, but that seems like a less-pressing issue; how many compilers are going to mess up simply because fd 5 is open? diff --git a/lib/autotest/general.m4 b/lib/autotest/general.m4 index 78f1cc6..fc8d866 100644 --- a/lib/autotest/general.m4 +++ b/lib/autotest/general.m4 @@ -2190,7 +2190,7 @@ m4_define([_AT_CHECK], AS_ECHO(["$at_srcdir/AT_LINE: AS_ESCAPE([[$1]])"]) _AT_DECIDE_TRACEABLE([$1]) _AT_LINE_ESCAPED ( $at_check_trace; [$1] -) >>"$at_stdout" 2>>"$at_stderr" +) >>"$at_stdout" 2>>"$at_stderr" AS_MESSAGE_LOG_FD>&- at_status=$? at_failed=false $at_check_filter m4_ifdef([AT_DIFF_STDERR($4)], [m4_indir([AT_DIFF_STDERR($4)])], _______________________________________________ Autoconf mailing list Autoconf@xxxxxxx http://lists.gnu.org/mailman/listinfo/autoconf
