On Thu, Mar 12, 2020 at 9:22 PM Paul Eggert <egg...@cs.ucla.edu> wrote: > On 3/12/20 1:53 PM, Zack Weinberg wrote: > > the > > changed behavior of autoconf trunk was not intentional; it's because > > commit 2b59b6f8a79b8bf77e178ff4e5aa0ede433d39cf missed a direct use of > > $as_echo in _AS_DETECT_BETTER_SHELL, causing _AS_DETECT_BETTER_SHELL > > *not to work at all*! > > Thanks for catching that. I looked into the history of that part of > Autoconf and it appears the string is being echoed only for historical > reasons. So I installed the attached to bypass the need for using > $as_echo or equivalent.
Thanks for taking care of that. I wouldn't have been bold enough to switch to sh -c. Out of curiosity, what was the historical reason for echoing the string into the shell? FYI, I grepped the whole autoconf source tree and I didn't find any other uses of $as_echo. However, there is one use in libtool/tests/testsuite.at and one use in autoconf-archive/m4/ax_pthread.m4. I'm going to send patches to both of those projects, but the use in ax_pthread.m4 makes me think we might need to provide $as_echo and $as_echo_n for backward compatibility; people copy stuff out of the autoconf archive and then don't ever look for updates. Also, this bug having gone unnoticed since 2013 says to me that we need more tests for _AS_DETECT_BETTER_SHELL. Would you happen to have time to write some? The m4sh layer is mostly a black box to me.. zw
