>>>>> "PR" == Pavel Roskin <[EMAIL PROTECTED]> writes:
PR> Hello, Martin!
>> Let's think about when you would ever want to reuse a cache within a
>> simple single GNU package. I claim you _never_ want to reuse that
>> config.cache file sitting in your foo-1.9 directory. Why are you
>> re-running configure, anyways, instead of just running `make'?
PR> Because I want to try a different _configuration_:
PR> CFLAGS=-ggdb3 ./configure --enable-cute-feature --without-bloat
If all you want is a debugging version, you should be able to do
make clean; make CFLAGS=-ggdb3
I do this myself all the time. But I also keep at the back of my mind
that it's my knowledge of the particular compiler and flags involved
that this is mostly safe. The configure script itself cannot affort
to be so reckless.
If there's a reason to rerun configure in this case just for CFLAGS,
then there's a reason to actually rerun all the tests.
PR> Cache really saves time when I turn on/off debugging and optional
PR> features. It remains valid in this case.
It _might_ work.
If you do --with-x11, presumably configure will play tricks with -I
and -L flags. In this case, all the compilation and link tests need
to be retried, if you want to be safe.
It's true that your example points out two mostly orthogonal purposes
of configure - condensing the environment into an environment
description, and recording user choices.
