On 08/21/2010 01:42 AM, Ralf Wildenhues wrote:
Any hints as to what could make the tests fail when run as part of the M4
git tree and not otherwise?
I don't envy you debugging that, but I have seen similar discrepancies
for directory-style tests when my two source trees are in different file
systems; perhaps that might explain your rmdir results. For the strtod
stuff the HAVE_DECL_STRTOD discrepancy is odd; perhaps you can look in
the logs and see why that is.
I never bother using older autoconfs, so I can't be of much help with
anything older than autoconf 2.67. If an upgrade to that version fixes
the problem, then evidently there would be some use in gnulib that
assumes newer autoconf.
