compound checks (was: AC_REQUIRE problems)

[Stepan Kasal]

Hello Dan,
  I'll try to answer; it's quite possible that Paul will correct me.

On Fri, Feb 18, 2005 at 11:55:27AM -0500, Dan Manthey wrote:
> checking for a type like uint_fast32_t... (This may take a while.)
> checking for blah...
> ...
> checking finished for a type like uint_fast32_t... found

the basic philosphy of autoconf is that you run _small_, mostly
_independent_ tests.  (That's why Paul suggested make, a *static*
dependency resolution doesn't sound that unreasonable.)

So the right solution should probably be:

> checking for blah...
> ...
> checking finished for a type like uint_fast32_t... found

ie. all the individual tests are listed and then in one line for the final
decision.

If the values are cached, then all the individual tests are cached, so the
output looks the same, but each line contains the note: "(cached)".

If you want to mark that group of tests, you can use AC_MSG_NOTICE.
Let me cite from the manual:

 - Macro: AC_MSG_NOTICE (MESSAGE)
     Deliver the MESSAGE to the user.  It is useful mainly to print a
     general description of the overall purpose of a group of feature
     checks, e.g.,
          AC_MSG_NOTICE([checking if stack overflow is detectable])

I, personally, think that the lines "checking for ..." should only mark
non-trivial tests, which involve running a tool, eg. a compiler.
So if there is some simple logic which decides for uint_fast32_t as soon
as the particular checks are performed, it's perfectly possible that
I'd omit the last line (the one containing:
        checking for a type like uint_fast32_t... found
).

HTH,
        Stepan