Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[tlaronde]

On Tue, Nov 14, 2017 at 08:12:20PM -0500, Eric Haszlakiewicz wrote:
> On Nov 14, 2017 10:39 AM,  wrote:
> 
> On Tue, Nov 14, 2017 at 03:08:29PM +0100, Benny Siegert wrote:
> > Can you post the entire config.log?
> >
> 
> Attached.
> 
> (Indeed there is the info: libssp not found---what is it?)
> 
> 
> You might find some interesting reading over at:
> 
> https://wiki.netbsd.org/pkgsrc/hardening/
> 

Thank you for the pointer!
-- 
Thierry Laronde 
 http://www.kergis.com/
   http://www.sbfa.fr/
Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[tlaronde]

On Wed, Nov 15, 2017 at 12:50:18AM +, m...@netbsd.org wrote:
> On Tue, Nov 14, 2017 at 03:21:47PM +0100, tlaro...@polynum.com wrote:
> > On Tue, Nov 14, 2017 at 03:08:29PM +0100, Benny Siegert wrote:
> > > Can you post the entire config.log?
> > > 
> > 
> > Attached.
> > 
> > (Indeed there is the info: libssp not found---what is it?)
> > 
> 
> Hi,
> 
> SSP is a compiler security mitigation.
> In C are some functions which are more prone to buffer overflows, and
> while there are better functions to use, they still exist in a lot of
> code (and are easier to use).
> 
> I think the compiler inserts fake values to the end of buffers and
> checks if it has been modified, and so it can terminate the program if
> an overflow had happened (as opposed to possibly allowing code
> execution).
> 
> netbsd did not want to supply libssp, an independent library from GCC,
> and so provides the same in libc. GCC had to be taught not to build
> libssp on netbsd, but didn't in the past.
> 
> The problem you have is likely an older gcc48/49 package that does not
> contain the fixes not to think netbsd built with -lssp and
> -lssp_nonshared.  Others have fixed some more changes.

Yes, I have a gcc48 package built long ago. To be thorough, the gcc
shipped with NetBSD 6.1.5 is in the same case.

Thanks for the info!
-- 
Thierry Laronde 
 http://www.kergis.com/
   http://www.sbfa.fr/
Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[Eric Haszlakiewicz]

On Nov 14, 2017 10:39 AM,  wrote:

On Tue, Nov 14, 2017 at 03:08:29PM +0100, Benny Siegert wrote:
> Can you post the entire config.log?
>

Attached.

(Indeed there is the info: libssp not found---what is it?)


You might find some interesting reading over at:

https://wiki.netbsd.org/pkgsrc/hardening/

Eric

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[maya]

On Tue, Nov 14, 2017 at 03:21:47PM +0100, tlaro...@polynum.com wrote:
> On Tue, Nov 14, 2017 at 03:08:29PM +0100, Benny Siegert wrote:
> > Can you post the entire config.log?
> > 
> 
> Attached.
> 
> (Indeed there is the info: libssp not found---what is it?)
> 

Hi,

SSP is a compiler security mitigation.
In C are some functions which are more prone to buffer overflows, and
while there are better functions to use, they still exist in a lot of
code (and are easier to use).

I think the compiler inserts fake values to the end of buffers and
checks if it has been modified, and so it can terminate the program if
an overflow had happened (as opposed to possibly allowing code
execution).

netbsd did not want to supply libssp, an independent library from GCC,
and so provides the same in libc. GCC had to be taught not to build
libssp on netbsd, but didn't in the past.

The problem you have is likely an older gcc48/49 package that does not
contain the fixes not to think netbsd built with -lssp and
-lssp_nonshared.  Others have fixed some more changes.

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[tlaronde]

On Tue, Nov 14, 2017 at 07:56:14PM +0100, Benny Siegert wrote:
> > But for my information, libssp is linked (or not) with gcc but what is
> > it supposed to provide?
> 
> Stack Smashing Protection. It is a security feature.

Thank you to both of you. Problem closed.
-- 
Thierry Laronde 
 http://www.kergis.com/
   http://www.sbfa.fr/
Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[Benny Siegert]

> But for my information, libssp is linked (or not) with gcc but what is
> it supposed to provide?

Stack Smashing Protection. It is a security feature.

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[tlaronde]

Hello,

On Tue, Nov 14, 2017 at 06:26:09PM +, Chavdar Ivanov wrote:
> PKGSRC_USE_SSP=NO in /etc/mk.conf
> 

Thank you! This does the trick.

But for my information, libssp is linked (or not) with gcc but what is
it supposed to provide?

Best regards.

T. Laronde
> 
> On 14 November 2017 at 14:21,   wrote:
> > On Tue, Nov 14, 2017 at 03:08:29PM +0100, Benny Siegert wrote:
> >> Can you post the entire config.log?
> >>
> >
> > Attached.
> >
> > (Indeed there is the info: libssp not found---what is it?)
> >
> > Best regards.
> >
> >> On Tue, Nov 14, 2017 at 3:06 PM,   wrote:
> >> > On Tue, Nov 14, 2017 at 02:04:44PM +0100, Benny Siegert wrote:
> >> >> > When configure is run from pkgsrc, it fails.
> >> >> >
> >> >> > Not knowing the guts of the pkgsrc framework, I'm a bit at a loss to
> >> >> > have a clue about what is going wrong...
> >> >>
> >> >> Please do the following:
> >> >>
> >> >> 1. Run "make configure", watch it fail.
> >> >> 2. Look in work/icu-*/config.log for the failing compiler invocation.
> >> >> Post the relevant bit.
> >> >
> >> > The only info in config.log is:
> >> >
> >> > configure: exit 77
> >> >
> >> > The relevant bit failing is the invocation of the compiler to find the
> >> > extension (suffix) of the executable, that is (extract from configure):
> >> >
> >> > 8<---
> >> > cat confdefs.h - <<_ACEOF >conftest.$ac_ext
> >> > /* end confdefs.h.  */
> >> >
> >> > int
> >> > main ()
> >> > {
> >> >
> >> >   ;
> >> >   return 0;
> >> > }
> >> > _ACEOF
> >> > ac_clean_files_save=$ac_clean_files
> >> > ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
> >> > # Try to create an executable without -o first, disregard a.out.
> >> > # It will help us diagnose broken compilers, and finding out an intuition
> >> > # of exeext.
> >> > { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C compiler 
> >> > works" >&5
> >> > $as_echo_n "checking whether the C compiler works... " >&6; }
> >> > ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
> >> >
> >> > # The possible output files:
> >> > ac_files="a.out conftest.exe conftest a.exe a_out.exe b.out conftest.*"
> >> >
> >> > ac_rmfiles=
> >> > for ac_file in $ac_files
> >> > do
> >> >   case $ac_file in
> >> > *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | 
> >> > *.map | *.inf | *.dSYM | *.o | *.obj ) ;;
> >> > * ) ac_rmfiles="$ac_rmfiles $ac_file";;
> >> >   esac
> >> > done
> >> > rm -f $ac_rmfiles
> >> >
> >> > if { { ac_try="$ac_link_default"
> >> > case "(($ac_try" in
> >> >   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
> >> >   *) ac_try_echo=$ac_try;;
> >> > esac
> >> > eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
> >> > $as_echo "$ac_try_echo"; } >&5
> >> >   (eval "$ac_link_default") 2>&5
> >> >   ac_status=$?
> >> >   $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
> >> >   test $ac_status = 0; }; then :
> >> >   # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
> >> > # So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
> >> > # in a Makefile.  We should not override ac_cv_exeext if it was cached,
> >> > # so that the user can short-circuit this test for compilers unknown to
> >> > # Autoconf.
> >> > for ac_file in $ac_files ''
> >> > do
> >> >   test -f "$ac_file" || continue
> >> >   case $ac_file in
> >> > *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | 
> >> > *.map | *.inf | *.dSYM | *.o | *.obj )
> >> > ;;
> >> > [ab].out )
> >> > # We found the default executable, but exeext='' is most
> >> > # certainly right.
> >> > break;;
> >> > *.* )
> >> > if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != 
> >> > no;
> >> > then :; else
> >> >ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
> >> > fi
> >> > # We set ac_cv_exeext here because the later test for it is not
> >> > # safe: cross compilers may not add the suffix if given an `-o'
> >> > # argument, so we may need to know it at that point already.
> >> > # Even if this section looks crufty: it has the advantage of
> >> > # actually working.
> >> > break;;
> >> > * )
> >> > break;;
> >> >   esac
> >> > done
> >> > test "$ac_cv_exeext" = no && ac_cv_exeext=
> >> >
> >> > else
> >> >   ac_file=''
> >> > fi
> >> > if test -z "$ac_file"; then :
> >> >   { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
> >> > $as_echo "no" >&6; }
> >> > $as_echo "$as_me: failed program was:" >&5
> >> > sed 's/^/| /' conftest.$ac_ext >&5
> >> >
> >> > { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
> >> > $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
> >> > as_fn_error 77 "C compiler cannot create executables
> >> > See \`config.log' for more details" "$LINENO" 5; }
> >> > else
> >> >   { $as_echo 

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[Chavdar Ivanov]

PKGSRC_USE_SSP=NO in /etc/mk.conf

Chavdar Ivanov

On 14 November 2017 at 14:21,   wrote:
> On Tue, Nov 14, 2017 at 03:08:29PM +0100, Benny Siegert wrote:
>> Can you post the entire config.log?
>>
>
> Attached.
>
> (Indeed there is the info: libssp not found---what is it?)
>
> Best regards.
>
>> On Tue, Nov 14, 2017 at 3:06 PM,   wrote:
>> > On Tue, Nov 14, 2017 at 02:04:44PM +0100, Benny Siegert wrote:
>> >> > When configure is run from pkgsrc, it fails.
>> >> >
>> >> > Not knowing the guts of the pkgsrc framework, I'm a bit at a loss to
>> >> > have a clue about what is going wrong...
>> >>
>> >> Please do the following:
>> >>
>> >> 1. Run "make configure", watch it fail.
>> >> 2. Look in work/icu-*/config.log for the failing compiler invocation.
>> >> Post the relevant bit.
>> >
>> > The only info in config.log is:
>> >
>> > configure: exit 77
>> >
>> > The relevant bit failing is the invocation of the compiler to find the
>> > extension (suffix) of the executable, that is (extract from configure):
>> >
>> > 8<---
>> > cat confdefs.h - <<_ACEOF >conftest.$ac_ext
>> > /* end confdefs.h.  */
>> >
>> > int
>> > main ()
>> > {
>> >
>> >   ;
>> >   return 0;
>> > }
>> > _ACEOF
>> > ac_clean_files_save=$ac_clean_files
>> > ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
>> > # Try to create an executable without -o first, disregard a.out.
>> > # It will help us diagnose broken compilers, and finding out an intuition
>> > # of exeext.
>> > { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C compiler 
>> > works" >&5
>> > $as_echo_n "checking whether the C compiler works... " >&6; }
>> > ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
>> >
>> > # The possible output files:
>> > ac_files="a.out conftest.exe conftest a.exe a_out.exe b.out conftest.*"
>> >
>> > ac_rmfiles=
>> > for ac_file in $ac_files
>> > do
>> >   case $ac_file in
>> > *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | 
>> > *.map | *.inf | *.dSYM | *.o | *.obj ) ;;
>> > * ) ac_rmfiles="$ac_rmfiles $ac_file";;
>> >   esac
>> > done
>> > rm -f $ac_rmfiles
>> >
>> > if { { ac_try="$ac_link_default"
>> > case "(($ac_try" in
>> >   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
>> >   *) ac_try_echo=$ac_try;;
>> > esac
>> > eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
>> > $as_echo "$ac_try_echo"; } >&5
>> >   (eval "$ac_link_default") 2>&5
>> >   ac_status=$?
>> >   $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
>> >   test $ac_status = 0; }; then :
>> >   # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
>> > # So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
>> > # in a Makefile.  We should not override ac_cv_exeext if it was cached,
>> > # so that the user can short-circuit this test for compilers unknown to
>> > # Autoconf.
>> > for ac_file in $ac_files ''
>> > do
>> >   test -f "$ac_file" || continue
>> >   case $ac_file in
>> > *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | 
>> > *.map | *.inf | *.dSYM | *.o | *.obj )
>> > ;;
>> > [ab].out )
>> > # We found the default executable, but exeext='' is most
>> > # certainly right.
>> > break;;
>> > *.* )
>> > if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
>> > then :; else
>> >ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
>> > fi
>> > # We set ac_cv_exeext here because the later test for it is not
>> > # safe: cross compilers may not add the suffix if given an `-o'
>> > # argument, so we may need to know it at that point already.
>> > # Even if this section looks crufty: it has the advantage of
>> > # actually working.
>> > break;;
>> > * )
>> > break;;
>> >   esac
>> > done
>> > test "$ac_cv_exeext" = no && ac_cv_exeext=
>> >
>> > else
>> >   ac_file=''
>> > fi
>> > if test -z "$ac_file"; then :
>> >   { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
>> > $as_echo "no" >&6; }
>> > $as_echo "$as_me: failed program was:" >&5
>> > sed 's/^/| /' conftest.$ac_ext >&5
>> >
>> > { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
>> > $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
>> > as_fn_error 77 "C compiler cannot create executables
>> > See \`config.log' for more details" "$LINENO" 5; }
>> > else
>> >   { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
>> > $as_echo "yes" >&6; }
>> > fi
>> > --->8---
>> >
>> > Since this is this configure, produced by pkgsrc, that I run directly,
>> > except the way it is invoked (and the directory), I fail to grab what
>> > may be causing the problem---unless the environment set by make(1), with
>> > the numerous utilities set by macros, replace a default working by a not
>> > working one.
>> >
>> > TIA,
>> > --
>> > Thierry Laronde 
>> >  

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[tlaronde]

On Tue, Nov 14, 2017 at 03:08:29PM +0100, Benny Siegert wrote:
> Can you post the entire config.log?
> 

Attached.

(Indeed there is the info: libssp not found---what is it?)

Best regards.

> On Tue, Nov 14, 2017 at 3:06 PM,   wrote:
> > On Tue, Nov 14, 2017 at 02:04:44PM +0100, Benny Siegert wrote:
> >> > When configure is run from pkgsrc, it fails.
> >> >
> >> > Not knowing the guts of the pkgsrc framework, I'm a bit at a loss to
> >> > have a clue about what is going wrong...
> >>
> >> Please do the following:
> >>
> >> 1. Run "make configure", watch it fail.
> >> 2. Look in work/icu-*/config.log for the failing compiler invocation.
> >> Post the relevant bit.
> >
> > The only info in config.log is:
> >
> > configure: exit 77
> >
> > The relevant bit failing is the invocation of the compiler to find the
> > extension (suffix) of the executable, that is (extract from configure):
> >
> > 8<---
> > cat confdefs.h - <<_ACEOF >conftest.$ac_ext
> > /* end confdefs.h.  */
> >
> > int
> > main ()
> > {
> >
> >   ;
> >   return 0;
> > }
> > _ACEOF
> > ac_clean_files_save=$ac_clean_files
> > ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
> > # Try to create an executable without -o first, disregard a.out.
> > # It will help us diagnose broken compilers, and finding out an intuition
> > # of exeext.
> > { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C compiler 
> > works" >&5
> > $as_echo_n "checking whether the C compiler works... " >&6; }
> > ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
> >
> > # The possible output files:
> > ac_files="a.out conftest.exe conftest a.exe a_out.exe b.out conftest.*"
> >
> > ac_rmfiles=
> > for ac_file in $ac_files
> > do
> >   case $ac_file in
> > *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | 
> > *.map | *.inf | *.dSYM | *.o | *.obj ) ;;
> > * ) ac_rmfiles="$ac_rmfiles $ac_file";;
> >   esac
> > done
> > rm -f $ac_rmfiles
> >
> > if { { ac_try="$ac_link_default"
> > case "(($ac_try" in
> >   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
> >   *) ac_try_echo=$ac_try;;
> > esac
> > eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
> > $as_echo "$ac_try_echo"; } >&5
> >   (eval "$ac_link_default") 2>&5
> >   ac_status=$?
> >   $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
> >   test $ac_status = 0; }; then :
> >   # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
> > # So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
> > # in a Makefile.  We should not override ac_cv_exeext if it was cached,
> > # so that the user can short-circuit this test for compilers unknown to
> > # Autoconf.
> > for ac_file in $ac_files ''
> > do
> >   test -f "$ac_file" || continue
> >   case $ac_file in
> > *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | 
> > *.map | *.inf | *.dSYM | *.o | *.obj )
> > ;;
> > [ab].out )
> > # We found the default executable, but exeext='' is most
> > # certainly right.
> > break;;
> > *.* )
> > if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
> > then :; else
> >ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
> > fi
> > # We set ac_cv_exeext here because the later test for it is not
> > # safe: cross compilers may not add the suffix if given an `-o'
> > # argument, so we may need to know it at that point already.
> > # Even if this section looks crufty: it has the advantage of
> > # actually working.
> > break;;
> > * )
> > break;;
> >   esac
> > done
> > test "$ac_cv_exeext" = no && ac_cv_exeext=
> >
> > else
> >   ac_file=''
> > fi
> > if test -z "$ac_file"; then :
> >   { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
> > $as_echo "no" >&6; }
> > $as_echo "$as_me: failed program was:" >&5
> > sed 's/^/| /' conftest.$ac_ext >&5
> >
> > { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
> > $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
> > as_fn_error 77 "C compiler cannot create executables
> > See \`config.log' for more details" "$LINENO" 5; }
> > else
> >   { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
> > $as_echo "yes" >&6; }
> > fi
> > --->8---
> >
> > Since this is this configure, produced by pkgsrc, that I run directly,
> > except the way it is invoked (and the directory), I fail to grab what
> > may be causing the problem---unless the environment set by make(1), with
> > the numerous utilities set by macros, replace a default working by a not
> > working one.
> >
> > TIA,
> > --
> > Thierry Laronde 
> >  http://www.kergis.com/
> >http://www.sbfa.fr/
> > Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C
> 
> 
> 
> -- 
> The first essential in chemistry is that you should perform practical
> work and conduct 

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[tlaronde]

On Tue, Nov 14, 2017 at 02:04:44PM +0100, Benny Siegert wrote:
> > When configure is run from pkgsrc, it fails.
> >
> > Not knowing the guts of the pkgsrc framework, I'm a bit at a loss to
> > have a clue about what is going wrong...
> 
> Please do the following:
> 
> 1. Run "make configure", watch it fail.
> 2. Look in work/icu-*/config.log for the failing compiler invocation.
> Post the relevant bit.

The only info in config.log is:

configure: exit 77

The relevant bit failing is the invocation of the compiler to find the
extension (suffix) of the executable, that is (extract from configure):

8<---
cat confdefs.h - <<_ACEOF >conftest.$ac_ext
/* end confdefs.h.  */

int
main ()
{

  ;
  return 0;
}
_ACEOF
ac_clean_files_save=$ac_clean_files
ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
# Try to create an executable without -o first, disregard a.out.
# It will help us diagnose broken compilers, and finding out an intuition
# of exeext.
{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C compiler works" 
>&5
$as_echo_n "checking whether the C compiler works... " >&6; }
ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`

# The possible output files:
ac_files="a.out conftest.exe conftest a.exe a_out.exe b.out conftest.*"

ac_rmfiles=
for ac_file in $ac_files
do
  case $ac_file in
*.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | 
*.inf | *.dSYM | *.o | *.obj ) ;;
* ) ac_rmfiles="$ac_rmfiles $ac_file";;
  esac
done
rm -f $ac_rmfiles

if { { ac_try="$ac_link_default"
case "(($ac_try" in
  *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
  *) ac_try_echo=$ac_try;;
esac
eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
$as_echo "$ac_try_echo"; } >&5
  (eval "$ac_link_default") 2>&5
  ac_status=$?
  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
  test $ac_status = 0; }; then :
  # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
# So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
# in a Makefile.  We should not override ac_cv_exeext if it was cached,
# so that the user can short-circuit this test for compilers unknown to
# Autoconf.
for ac_file in $ac_files ''
do
  test -f "$ac_file" || continue
  case $ac_file in
*.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map | 
*.inf | *.dSYM | *.o | *.obj )
;;
[ab].out )
# We found the default executable, but exeext='' is most
# certainly right.
break;;
*.* )
if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
then :; else
   ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
fi
# We set ac_cv_exeext here because the later test for it is not
# safe: cross compilers may not add the suffix if given an `-o'
# argument, so we may need to know it at that point already.
# Even if this section looks crufty: it has the advantage of
# actually working.
break;;
* )
break;;
  esac
done
test "$ac_cv_exeext" = no && ac_cv_exeext=

else
  ac_file=''
fi
if test -z "$ac_file"; then :
  { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
$as_echo "no" >&6; }
$as_echo "$as_me: failed program was:" >&5
sed 's/^/| /' conftest.$ac_ext >&5

{ { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
$as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
as_fn_error 77 "C compiler cannot create executables
See \`config.log' for more details" "$LINENO" 5; }
else
  { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
$as_echo "yes" >&6; }
fi
--->8---

Since this is this configure, produced by pkgsrc, that I run directly,
except the way it is invoked (and the directory), I fail to grab what
may be causing the problem---unless the environment set by make(1), with
the numerous utilities set by macros, replace a default working by a not
working one.

TIA,
-- 
Thierry Laronde 
 http://www.kergis.com/
   http://www.sbfa.fr/
Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[Benny Siegert]

Can you post the entire config.log?

On Tue, Nov 14, 2017 at 3:06 PM,   wrote:
> On Tue, Nov 14, 2017 at 02:04:44PM +0100, Benny Siegert wrote:
>> > When configure is run from pkgsrc, it fails.
>> >
>> > Not knowing the guts of the pkgsrc framework, I'm a bit at a loss to
>> > have a clue about what is going wrong...
>>
>> Please do the following:
>>
>> 1. Run "make configure", watch it fail.
>> 2. Look in work/icu-*/config.log for the failing compiler invocation.
>> Post the relevant bit.
>
> The only info in config.log is:
>
> configure: exit 77
>
> The relevant bit failing is the invocation of the compiler to find the
> extension (suffix) of the executable, that is (extract from configure):
>
> 8<---
> cat confdefs.h - <<_ACEOF >conftest.$ac_ext
> /* end confdefs.h.  */
>
> int
> main ()
> {
>
>   ;
>   return 0;
> }
> _ACEOF
> ac_clean_files_save=$ac_clean_files
> ac_clean_files="$ac_clean_files a.out a.out.dSYM a.exe b.out"
> # Try to create an executable without -o first, disregard a.out.
> # It will help us diagnose broken compilers, and finding out an intuition
> # of exeext.
> { $as_echo "$as_me:${as_lineno-$LINENO}: checking whether the C compiler 
> works" >&5
> $as_echo_n "checking whether the C compiler works... " >&6; }
> ac_link_default=`$as_echo "$ac_link" | sed 's/ -o *conftest[^ ]*//'`
>
> # The possible output files:
> ac_files="a.out conftest.exe conftest a.exe a_out.exe b.out conftest.*"
>
> ac_rmfiles=
> for ac_file in $ac_files
> do
>   case $ac_file in
> *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map 
> | *.inf | *.dSYM | *.o | *.obj ) ;;
> * ) ac_rmfiles="$ac_rmfiles $ac_file";;
>   esac
> done
> rm -f $ac_rmfiles
>
> if { { ac_try="$ac_link_default"
> case "(($ac_try" in
>   *\"* | *\`* | *\\*) ac_try_echo=\$ac_try;;
>   *) ac_try_echo=$ac_try;;
> esac
> eval ac_try_echo="\"\$as_me:${as_lineno-$LINENO}: $ac_try_echo\""
> $as_echo "$ac_try_echo"; } >&5
>   (eval "$ac_link_default") 2>&5
>   ac_status=$?
>   $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
>   test $ac_status = 0; }; then :
>   # Autoconf-2.13 could set the ac_cv_exeext variable to `no'.
> # So ignore a value of `no', otherwise this would lead to `EXEEXT = no'
> # in a Makefile.  We should not override ac_cv_exeext if it was cached,
> # so that the user can short-circuit this test for compilers unknown to
> # Autoconf.
> for ac_file in $ac_files ''
> do
>   test -f "$ac_file" || continue
>   case $ac_file in
> *.$ac_ext | *.xcoff | *.tds | *.d | *.pdb | *.xSYM | *.bb | *.bbg | *.map 
> | *.inf | *.dSYM | *.o | *.obj )
> ;;
> [ab].out )
> # We found the default executable, but exeext='' is most
> # certainly right.
> break;;
> *.* )
> if test "${ac_cv_exeext+set}" = set && test "$ac_cv_exeext" != no;
> then :; else
>ac_cv_exeext=`expr "$ac_file" : '[^.]*\(\..*\)'`
> fi
> # We set ac_cv_exeext here because the later test for it is not
> # safe: cross compilers may not add the suffix if given an `-o'
> # argument, so we may need to know it at that point already.
> # Even if this section looks crufty: it has the advantage of
> # actually working.
> break;;
> * )
> break;;
>   esac
> done
> test "$ac_cv_exeext" = no && ac_cv_exeext=
>
> else
>   ac_file=''
> fi
> if test -z "$ac_file"; then :
>   { $as_echo "$as_me:${as_lineno-$LINENO}: result: no" >&5
> $as_echo "no" >&6; }
> $as_echo "$as_me: failed program was:" >&5
> sed 's/^/| /' conftest.$ac_ext >&5
>
> { { $as_echo "$as_me:${as_lineno-$LINENO}: error: in \`$ac_pwd':" >&5
> $as_echo "$as_me: error: in \`$ac_pwd':" >&2;}
> as_fn_error 77 "C compiler cannot create executables
> See \`config.log' for more details" "$LINENO" 5; }
> else
>   { $as_echo "$as_me:${as_lineno-$LINENO}: result: yes" >&5
> $as_echo "yes" >&6; }
> fi
> --->8---
>
> Since this is this configure, produced by pkgsrc, that I run directly,
> except the way it is invoked (and the directory), I fail to grab what
> may be causing the problem---unless the environment set by make(1), with
> the numerous utilities set by macros, replace a default working by a not
> working one.
>
> TIA,
> --
> Thierry Laronde 
>  http://www.kergis.com/
>http://www.sbfa.fr/
> Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C



-- 
The first essential in chemistry is that you should perform practical
work and conduct experiments, for he who performs not practical work
nor makes experiments will never attain the least degree of mastery.
-- Abu Musa Jabir ibn Hayyan (721-815)

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[Benny Siegert]

> When configure is run from pkgsrc, it fails.
>
> Not knowing the guts of the pkgsrc framework, I'm a bit at a loss to
> have a clue about what is going wrong...

Please do the following:

1. Run "make configure", watch it fail.
2. Look in work/icu-*/config.log for the failing compiler invocation.
Post the relevant bit.

Thanks!

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[tlaronde]

Hello and thank you for the answer,

On Tue, Nov 14, 2017 at 01:36:42PM +0100, Benny Siegert wrote:
> Is that a zero or an uppercase O?
> 

This is a zero but I found that this has nothing to do with the issue
(see below).

> This sounds very much like a local issue. Can you look in the
> config.log of a failed configure run what the command line and the
> error was? (I assume you are overriding something else in your default
> CFLAGS with your call.)
> 

In fact, if I run the configure in the ICU work directory with all the
args echoed at the beginning of the config.log, it works. I mean running
configure in pkgsrc/textproc/icu/work/icu/source/, it works.

When configure is run from pkgsrc, it fails.

Not knowing the guts of the pkgsrc framework, I'm a bit at a loss to
have a clue about what is going wrong...

TIA,
-- 
Thierry Laronde 
 http://www.kergis.com/
   http://www.sbfa.fr/
Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C

Re: [pkgsrc] ICU needs configure CFLAGS='-03'

[Benny Siegert]

Is that a zero or an uppercase O?

This sounds very much like a local issue. Can you look in the
config.log of a failed configure run what the command line and the
error was? (I assume you are overriding something else in your default
CFLAGS with your call.)

-- 
Benny

On Tue, Nov 14, 2017 at 11:04 AM,   wrote:
> Hello,
>
> As a dependency to firefox, ICU needs to be recompiled.
>
> I'm on:
>
> NetBSD 6.1.5_PATCH amd64
>
> This is pkgsrc-2017Q3.
>
> The ICU fails during configure with "C compiler cannot create
> executables". I've found on the net that invoking configure with:
>
> ./configure CFLAGS='-03'
>
> does the trick.
>
> But I don't understand how autoconf and al. work and I don't
> understand either how the pkgsrc framework invokes configure.
>
> Trying to set whether CONFIGURE_ARGS or adding to CFLAGS doesn't do it.
>
> Does anybody know how to instruct "pkgsrc" to set this or,
> alternatively, what is the name of the sentry file to put in the
> work directory after invoking configure by end so that pkgsrc will
> proceed?
>
> TIA for any tip.
>
> Best,
> --
> Thierry Laronde 
>  http://www.kergis.com/
>http://www.sbfa.fr/
> Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C



-- 
The first essential in chemistry is that you should perform practical
work and conduct experiments, for he who performs not practical work
nor makes experiments will never attain the least degree of mastery.
-- Abu Musa Jabir ibn Hayyan (721-815)