This actually looks like a weakness that libffi doesn't explicitly support Clang [0]. Swimming further upstream, this support is missing in the Autoconf Archive [1]. Whether this is worth fixing is debatable as even the GCC options in this script don't seem particularly well chosen for a modern toolchain.
[0]: https://github.com/libffi/libffi/blob/master/m4/ax_cc_maxopt.m4#L73-L155 [1]: http://git.savannah.gnu.org/gitweb/?p=autoconf-archive.git;a=blob_plain;f=m4...
On 16/11/17 13:42, Rob Arthan wrote:
David,
On 15 Nov 2017, at 17:58, David Matthews <David.Matthews at prolingua.co.uk> wrote: I did a search and came up with this: https://www.gnu.org/software/autoconf/manual/autoconf-2.68/html_node/Option-... . Poly/ML's autoconf script uses AC_CONFIG_SUBDIRS to build libffi so that will disable all option checking. There doesn't seem to be any way to override this and turn it on within the script but --enable-option-checking does seem to work. However ./configure --enable-option-cheking doesn't produce any warning at all so it doesn't help if that is misspelled.
Thank! It's not as silly as it sounds:
alias conf="./configure --enable-option-checking"is the latest addition to my .bashrc file.
Regards,
Rob.
polyml mailing list polyml at inf.ed.ac.uk http://lists.inf.ed.ac.uk/mailman/listinfo/polyml