FESCo Proposal for blocking older version of autoconf & automake

Ralf Corsepius rc040203 at freenet.de
Wed May 7 04:53:10 UTC 2008

On Tue, 2008-05-06 at 11:15 -0500, Jason L Tibbitts III wrote:
> >>>>> "TK" == Toshio Kuratomi <a.badger at gmail.com> writes:
> TK> Despite incompatibilities, some packages will generate correct
> TK> Makefiles whether you use automake-1.4 or automake-1.9.
> I guess the real question, then, is how to tell that the shiny new
> automake version has produced an incorrect makefile.

> Are the failures
> obvious, or do you build it and hope that it doesn't fail silently
> later?
All of it. Fact is, automake breakdowns are _rare_, because automake is
pretty strict about its input files.

More common are cases where modern automakes complain and diagnose
issues in input files, older automakes silently had swallowed (Sometimes
these are real bugs, sometimes these are side-effects from automake
having tightened its syntax).

Much more critical is autoconf ... There have been quite a few cases of
it silently breaking "once known to work" configure scripts during
autoconf-2.13 -> autoconf-2.50 transition.


More information about the devel mailing list