autoconf/BUGS
Akim Demaille 7bf59957c4 acfunctions.m4 was still using the old AC_LIBOBJ_DECL.
Reported by Derek R. Price.
* tests/semantics.at (AC_REPLACE_FUNCS): New test.
* acfunctions.m4 (AC_REPLACE_FUNCS, _AC_LIBOBJ_ALLOCA): Use
AC_LIBSOURCES.
2001-02-05 17:33:00 +00:00

54 lines
1.6 KiB
Plaintext

-*- outline -*-
This file lists the bugs you must be aware of. Be sure to check this
file before using Autoconf, and especially CVS versions of Autoconf.
Autoconf must not be used in production if there are ``Serious'' bugs,
and use with caution an Autoconf with ``Important bugs''.
Many other bugs are registered on the GNATS server:
http://sources.redhat.com/cgi-bin/gnatsweb.pl?database=autoconf
Please, don't register bugs listed below: we already know we have to
address them.
* Status
/*-------------.
| Do not use. |
`-------------*/
* Serious Problems
** Broken compilers
configure does not properly diagnose broken compilers (because it
finds the failure before the code that checks whether the compiler
works). It is not to serious a bug, but most notably the test suite
will fail to handle gracefully missing compilers (typically the
Fortran compiler).
** Special characters and config.hin
The handling of special characters, most notable backslashes, of
course, might be nonportable in addition of being of clearly
specified.
** AC_TRY_EVAL of multi line commands
Is broken. The problem is mostly that, again, the policy wrt special
characters is not clearly defined. Because of this,
_AC_INIT_PREPARE_FS_SEPARATORS will output undesired messages, but is
not guilty.
./configure: /home/lrde/prof/akim/src/ace/tests; conftest.sh: Aucun fichier ou répertoire de ce type
AC_TRY_COMMAND is.
** AC_CHECK_PROG
and all the macros that use it, are paying the change for test -f to
test -x to determine whether a candidate path is an executable: now we
have false positives with directories which, indeed, are test -x.