Re: Make -Wc,foo behave like -Xcompiler foo in link mode.

[Peter Rosin]

Hi Ralf,

Den 2009-09-06 11:44 skrev Ralf Wildenhues:
> Hi Peter,
>
> * Peter Rosin wrote on Fri, Sep 04, 2009 at 10:31:14PM CEST:
> > Den 2009-09-04 19:44 skrev Ralf Wildenhues:
> > > Further, the patch needs a testsuite addition, but I can do that later
> > > if you don't want to.  I was thinking of a new set of tests flags.at and
> > > running a bunch of
> > >  libtool -n --mode=X ... $flag-foo
> > >
> > > for $flag in -Wl, -Wc, '-Xcompiler ' and '-Xlinker ', both library and
> > > program output, and grepping the output for ${wl}-foo and similar (I
> > > don't remember whether that was general enough for all compilers, you
> > > know w32 better).
> > Sigh, tests. I'd rather not, sorry. Maybe later, I hope someone
> > beats me to it though...
>
> Proposed patch below.  Comments before I apply?

Thanks for letting me off the hook :-) Comments inline and below
patch...

> Thanks,
> Ralf
>
>     Add testsuite exposure for passing of compiler and linker flags.
>     
>     * tests/flags.at (passing flags through libtool): New file, new
>     test, for bug fixed in previous commit.
>     * Makefile.am (TESTSUITE_AT): Add tests/flags.at.
>
> diff --git a/Makefile.am b/Makefile.am
> index a18955e..e22fcc2 100644
> --- a/Makefile.am
> +++ b/Makefile.am
> @@ -455,6 +455,7 @@ TESTSUITE_AT        = tests/testsuite.at \
>                   tests/duplicate_members.at \
>                   tests/duplicate_conv.at \
>                   tests/duplicate_deps.at \
> +                 tests/flags.at \
>                   tests/inherited_flags.at \
>                   tests/convenience.at \
>                   tests/link-order.at \
> diff --git a/tests/flags.at b/tests/flags.at
> new file mode 100644
> index 0000000..3de1fcb
> --- /dev/null
> +++ b/tests/flags.at
> @@ -0,0 +1,96 @@
> +# flags.at -- libtool passing of flags             -*- Autotest -*-
> +#
> +#   Copyright (C) 2009 Free Software Foundation, Inc.
> +#
> +#   This file is part of GNU Libtool.
> +#
> +# GNU Libtool is free software; you can redistribute it and/or
> +# modify it under the terms of the GNU General Public License as
> +# published by the Free Software Foundation; either version 2 of
> +# the License, or (at your option) any later version.
> +#
> +# GNU Libtool is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License
> +# along with GNU Libtool; see the file COPYING.  If not, a copy
> +# can be downloaded from  http://www.gnu.org/licenses/gpl.html,
> +# or obtained by writing to the Free Software Foundation, Inc.,
> +# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
> +####
> +
> +AT_SETUP([passing flags through libtool])
> +AT_KEYWORDS([libtool])
> +AT_KEYWORDS([CXX F77 FC])
> +
> +LDFLAGS="$LDFLAGS -no-undefined"
> +eval "`$LIBTOOL --config | $EGREP '^(FGREP)='`"
> +
> +AT_DATA([a.c],
> +[[int main () { return 0; }
> +]])


Maybe some platform will barf if you try to create a shared
library that exports the symbol 'main'? Maybe we'll just find
out when someone stumbles :-)


> +
> +AT_DATA([a.f],
> +[[      program main
> +      end
> +]])
> +
> +cp a.c a.cpp
> +cp a.f a.f90
> +
> +# Linker flags are not passed to the archiver, so don't test static libraries.
> +if $LIBTOOL --features | grep 'enable shared libraries'; then
> +  library_and_module='library.la "module.la -module -avoid-version"'
> +else
> +  library_and_module=
> +fi


Is it possible to have shared libraries for some tags but not
others? I honestly have no clue...


> +for tag in CC CXX F77 FC; do
> +  if $LIBTOOL --tag=$tag 2>&1 | grep 'unknown tag'; then
> +    continue
> +  fi
> +  case $tag in
> +  CC) compile="$CC $CPPFLAGS $CFLAGS" link="$CC $CFLAGS $LDFLAGS" source=a.c ;;
> +  CXX) compile="$CXX $CPPFLAGS $CXXFLAGS" link="$CXX $CXXFLAGS $LDFLAGS" 
> source=a.cpp;;
> +  F77) compile="$CC $CPPFLAGS $CFLAGS" link="$CC $CFLAGS $LDFLAGS" source=a.f 
> ;;
> +  FC) compile="$CC $CPPFLAGS $CFLAGS" link="$CC $CFLAGS $LDFLAGS" source=a.f90 
> ;;
> +  esac
> +
> +  eval "`$LIBTOOL --tag=$tag --config | $EGREP 
> '^(wl|archive_cmds|reload_cmds)='`"


You are not using $reload_cmds anywhere.


> +
> +  AT_CHECK([$LIBTOOL --tag=$tag --mode=compile $compile -c $source],
> +          [], [ignore], [ignore])
> +
> +  # Linker flags are prefixed with ${wl} iff they are passed to the
> +  # compiler driver, instead of directly to the linker.
> +  case $archive_cmds in
> +  *\$LD*\$linker_flags*) maybe_wl= ;;
> +  *) maybe_wl=$wl ;;
> +  esac
> +
> +  for flag in -Wc, -Wl, '-Xcompiler ' '-Xlinker '; do
> +    case $flag in
> +    -Wc, | -Xcompiler\ )
> +      AT_CHECK([$LIBTOOL -n --tag=$tag --mode=compile $compile ]dnl
> +              [$flag-foo -c $source], [], [stdout], [ignore])
> +      AT_CHECK([$FGREP " -foo" stdout], [], [ignore])
> +      flag_prefix=
> +      ;;
> +    -Wl, | -Xlinker\ )
> +      flag_prefix=$maybe_wl
> +      ;;
> +    esac
> +
> +    eval set program "$library_and_module"
> +    for output
> +    do
> +      AT_CHECK([$LIBTOOL -n --tag=$tag --mode=link $link ]dnl
> +              [-o $output a.lo -rpath /nowhere $flag-foo], [], [stdout], 
> [ignore])
> +      AT_CHECK([$FGREP " $flag_prefix-foo" stdout], [], [ignore])
> +    done
> +  done
> +done
> +
> +AT_CLEANUP

There needs to be some markers about what is being done when
there is a repetetive loop like this (48 times through the
innermost loop in the "worst case"). I remember hardship
trying to deduce exactly what failed when I had a failure
inside stresstest.at...

An example: If the following line fails, the test as written
doesn't give too many hints as to what it's trying to do:

../../tests/flags.at:89: $LIBTOOL -n --tag=$tag --mode=link $link              
-o $output a.lo -rpath /nowhere $flag-foo

Maybe it's as simple as using LT_AT_CHECK instead of AT_CHECK?

Cheers,
Peter