On 07/02/11 23:45, Brian Gough wrote:
At Sun, 6 Feb 2011 23:11:43 +0100,
Ralf Wildenhues wrote:
Back then, the consensus was to not make it the default because that was
deemed too dangerous for users. Various reports are cited, and also the
problem is mentioned that such kinds of failures tend to be quiet very
often and are hard to debug.
The reports are convincing for me that it would be bad to revive the
old behaviour.
Most of the GNU maintainers at FOSDEM didn't know about the -C option
though. Maybe it would help to publicise "-C" in the output so people
are more aware of it. e.g.
$ ./configure
configure: not caching test results (use -C to enable)
checking for a BSD-compatible install... /usr/bin/install -c
...
The message could be suppressed if config.status does not exist, so
it's only shown if someone is configuring more than once.
If you compile/install a program successfully, then apt-get remove a
library (or stuff gets uninstalled when installing something else),
then the manually compiled program doesn't run. A new
./configure will run successfully from the cached results even
though a required library is missing. Seems pretty bad to me if
using cached results is the default. Only developers doing
repeated ./configure need it, and can just use the -C option.
_______________________________________________
Autoconf mailing list
Autoconf@gnu.org
http://lists.gnu.org/mailman/listinfo/autoconf
