Eric Sunshine a dit: > On Sun, 28 Sep 2003 11:58:54 +0200 (CEST), Akim Demaille wrote: >> Could you make the test case smaller? Are you sure all the tests were >> run with the modified m4? > > Okay, I narrowed down the "excess arguments to m4_pushdef" to the > following > tiny case: > > AC_INIT([crystal], [1], [someone]) > AC_ARG_ENABLE([cpu-specific-optimizations], > [AC_HELP_STRING([--enable-cpu-specific-optimizations=level], > [foo, bar])], > [], []) > AC_OUTPUT > > It turns out that AC_HELP_STRING is the culprit. If I remove the comma > from > [foo, bar], then the error disappears. I suspect that the quoting in the > new definition of AC_HELP_STRING in 2.57d is not working correctly when it > calls the newly introduced AS_HELP_STRING. Arg... Now I understand why I didn't see the failure: I'm using CVS Autoconf, and AC_HELP_STRING has just been fixed since then :( I'm sorry for the waste of time... Still, is this the very same test case that causes other architectures to SEGV?
