Re: [PATCH] configure: Fix error message when C compiler is not working

[Thomas Huth]

On 19/03/2024 14.12, Peter Maydell wrote:
> On Fri, 8 Mar 2024 at 06:01, Thomas Huth <thuth@redhat.com> wrote:
> > If you try to run the configure script on a system without a working
> > C compiler, you get a very misleading error message:
> >
> >   ERROR: Unrecognized host OS (uname -s reports 'Linux')
> >
> > We should rather tell the user that we were not able to use the C
> > compiler instead, otherwise they will have a hard time to figure
> > out what was going wrong.
> >
> > Fixes: 264b803721 ("configure: remove compiler sanity check")
> > Signed-off-by: Thomas Huth <thuth@redhat.com>
> > ---
> >   configure | 12 +++++++-----
> >   1 file changed, 7 insertions(+), 5 deletions(-)
> >
> > diff --git a/configure b/configure
> > index 3cd736b139..a036923dee 100755
> > --- a/configure
> > +++ b/configure
> > @@ -411,7 +411,7 @@ else
> >     # Using uname is really broken, but it is just a fallback for architectures
> >     # that are going to use TCI anyway
> >     cpu=$(uname -m)
> > -  echo "WARNING: unrecognized host CPU, proceeding with 'uname -m' output 
> > '$cpu'"
> > +  echo "WARNING: could not determine host CPU, proceeding with 'uname -m' output 
> > '$cpu'"
> >   fi
> >
> >   # Normalise host CPU name to the values used by Meson cross files and in 
> > source
> > @@ -1000,10 +1000,12 @@ if test -z "$ninja"; then
> >   fi
> >
> >   if test "$host_os" = "bogus"; then
> > -    # Now that we know that we're not printing the help and that
> > -    # the compiler works (so the results of the check_defines we used
> > -    # to identify the OS are reliable), if we didn't recognize the
> > -    # host OS we should stop now.
> > +    # Now that we know that we're not printing the help, we should stop now
> > +    # if we didn't recognize the host OS (or the C compiler is not working).
> > +    write_c_skeleton;
> > +    if ! compile_object ; then
> > +        error_exit "C compiler \"$cc\" is not usable"
> > +    fi
> >       error_exit "Unrecognized host OS (uname -s reports '$(uname -s)')"
> >   fi
>
> I think I would prefer as a structure:
>
> (1) suppress the "unrecognized host CPU" message if "$host_os" == "bogus"
> (2) do the "check the C compiler works" test as its own test immediately
>      after we print the help message (and not guarded by testing $host_os)
> (3) leave the "Unrecognized host OS" check code as it is

Hmm, another idea: Why do we print the --help output that late in the 
configure script at all? Couldn't we move this earlier, to the place where 
we already check for the --cc et al. switches? Then we could get rid of the 
"bogus" stuff completely?

 Thomas