[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: AC_CACHE_CHECK
From: |
Ralf Wildenhues |
Subject: |
Re: AC_CACHE_CHECK |
Date: |
Sun, 5 Sep 2010 22:34:34 +0200 |
User-agent: |
Mutt/1.5.20 (2010-04-22) |
Hello,
* Thomas Dickey wrote on Sun, Sep 05, 2010 at 03:09:20PM CEST:
> On Sun, 5 Sep 2010, Russell Shaw wrote:
> >Thomas Dickey wrote:
> >>On Sun, 5 Sep 2010, Russell Shaw wrote:
> >>>When should AC_CACHE_CHECK be used? Whenever the user uninstalls
> >>>something, wouldn't the cache become invalid?
> >>
> >>...only if the user happened to uninstall something during the configure
> >>process. After that, AC_CACHE_CHECK is irrelevant.
> >
> >But if the user uninstalls libfoo then later re-runs ./configure,
> >won't libfoo still be assumed to be installed?
>
> sure. That sort of thing was why (for a long time) autoconf removed
> the cache on startup. Now that you're mentioning it, I seem to recall
> a recent undoing of that.
I can't remember that either autoconf nor configure would remove
config.cache, for any Autoconf >= 2.59. 'make distclean' from an
automake-generated Makefile does that.
But yes, you are supposed to not reuse the cache file after changing
your system.
With that in mind, AC_CACHE_CHECK should be used freely for any test
that is expensive or should be overridable by the user (because she
might know better), and that is amenable to single-entrance logic
(i.e., isn't run in a loop) and a unique cache variable name.
Cheers,
Ralf