RE: building gcc/g++ seems OK but building g77 seems to be seriously broken

RE: building gcc/g++ seems OK but building g77 seems to be seriously broken

[Billinghurst, David (CRTS)]

I would recommend installing installing dejagnu-1.4.1 (which you need for
libstdc++-v3) and fixing a line break problem
http://mail.gnu.org/pipermail/dejagnu/2001-May/000031.html

I posted some stuff about this to the cygwin, dejagnu and cygwin lists
earlier this year.  Try searching google using "cygwin dejagnu g77
billinghurst" for some background.



> -----Original Message-----
> From:	Mumit Khan [SMTP:khan@NanoTech.Wisc.EDU]
> Sent:	Wednesday, 18 July 2001 8:24
> To:	ted byers
> Cc:	cygwin@cygwin.com
> Subject:	Re: building gcc/g++ seems OK but building g77 seems to be
> seriously broken
> 
> On Tue, 17 Jul 2001, ted byers wrote:
> 
> > Well, I have now built GCC3 a couple times, trying to figure out what
> has
> > gone awry, but nothing seems to fix g77 (the summary given below show
> why I
> > think it is seriously broken).  gcc and g++ seem to be OK since their
> > respective test suites give the expected results (the number of
> unexpected
> > failures being extemely small).
> 
> When you report gcc build problems, you really need to provide the flags
> you passed to configure. Too many variables otherwise.
> 
> > 		=== g77 Summary ===
> > 
> > # of expected passes		281
> > # of unexpected failures	334
> > # of untested testcases		326
> > /home/Bted/Objects/gcc/g77 g77 version 3.0 (Fortran Frontend version
> 0.5.26
> > 20010617 (experimental))
> 
> Probably a problem in the g77 tests, which passes the -lm, and that causes
> a problem on Cygwin 1.1.x.
> 
> Regards,
> Mumit
> 
> 
> 
> --
> Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
> Bug reporting:         http://cygwin.com/bugs.html
> Documentation:         http://cygwin.com/docs.html
> FAQ:                   http://cygwin.com/faq/

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: building gcc/g++ seems OK but building g77 seems to be seriously broken

[ted byers]

Thanks,

I installed dejagnu when I installed cygwin, and I first did that only about
a couple of weeks ago.  Is that one not current, and wouldn't whoever
distributed the binaries for cygwin not have fixed the line break problem
you mention?

Thanks again,

Ted


-----Original Message-----
From: Billinghurst, David (CRTS)
[ mailto:David.Billinghurst@riotinto.com ]
Sent: Tuesday, July 17, 2001 7:56 PM
To: ted byers
Cc: cygwin@cygwin.com
Subject: RE: building gcc/g++ seems OK but building g77 seems to be
seriously broken


I would recommend installing installing dejagnu-1.4.1 (which you need for
libstdc++-v3) and fixing a line break problem
http://mail.gnu.org/pipermail/dejagnu/2001-May/000031.html

I posted some stuff about this to the cygwin, dejagnu and cygwin lists
earlier this year.  Try searching google using "cygwin dejagnu g77
billinghurst" for some background.




--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: building gcc/g++ seems OK but building g77 seems to be seriously broken

[ted byers]

Thanks,

I retrieved and installed dejagnu-1.4.1, and made the various edits
recommended (before proceeding with the installation).

I used the usual install procedure:

./configure
make
make install

However, it had no effect on the g77 tests, except that they took about 50%
longer to finish.

Cheers,

Ted

-----Original Message-----
From: Billinghurst, David (CRTS)
[ mailto:David.Billinghurst@riotinto.com ]
Sent: Tuesday, July 17, 2001 7:56 PM
To: ted byers
Cc: cygwin@cygwin.com
Subject: RE: building gcc/g++ seems OK but building g77 seems to be
seriously broken


I would recommend installing installing dejagnu-1.4.1 (which you need for
libstdc++-v3) and fixing a line break problem
http://mail.gnu.org/pipermail/dejagnu/2001-May/000031.html

I posted some stuff about this to the cygwin, dejagnu and cygwin lists
earlier this year.  Try searching google using "cygwin dejagnu g77
billinghurst" for some background.



> -----Original Message-----
> From:	Mumit Khan [SMTP:khan@NanoTech.Wisc.EDU]
> Sent:	Wednesday, 18 July 2001 8:24
> To:	ted byers
> Cc:	cygwin@cygwin.com
> Subject:	Re: building gcc/g++ seems OK but building g77 seems to be
> seriously broken
>
> On Tue, 17 Jul 2001, ted byers wrote:
>
> > Well, I have now built GCC3 a couple times, trying to figure out what
> has
> > gone awry, but nothing seems to fix g77 (the summary given below show
> why I
> > think it is seriously broken).  gcc and g++ seem to be OK since their
> > respective test suites give the expected results (the number of
> unexpected
> > failures being extemely small).
>
> When you report gcc build problems, you really need to provide the flags
> you passed to configure. Too many variables otherwise.
>
> > 		=== g77 Summary ===
> >
> > # of expected passes		281
> > # of unexpected failures	334
> > # of untested testcases		326
> > /home/Bted/Objects/gcc/g77 g77 version 3.0 (Fortran Frontend version
> 0.5.26
> > 20010617 (experimental))
>
> Probably a problem in the g77 tests, which passes the -lm, and that causes
> a problem on Cygwin 1.1.x.
>
> Regards,
> Mumit
>
>
>
> --
> Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
> Bug reporting:         http://cygwin.com/bugs.html
> Documentation:         http://cygwin.com/docs.html
> FAQ:                   http://cygwin.com/faq/


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: building gcc/g++ seems OK but building g77 seems to be seriously broken

[Charles Krug]

On Thu, Jul 19, 2001 at 09:13:48AM -0400, cygwin-return-33287-charles=pentek.com@sources.redhat.com wrote:
> Thanks,
> 
> I retrieved and installed dejagnu-1.4.1, and made the various edits
> recommended (before proceeding with the installation).
> 
> I used the usual install procedure:
> 
> ./configure
> make
> make install
> 

Don't you have to add a --with-languages or --enable-languages switch to
your command line to add FORTRAN support?



--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

RE: building gcc/g++ seems OK but building g77 seems to be seriously broken

[ted byers]

Thanks Mumit,

Oh OK,  The configure command I used was:

$
../Source/gcc-3.0/configure --prefix=/usr/local/gcc-3.0 --enable-threads --d
isable-win32-registry --enable-sjlj-exceptions --with-included-gettext --ena
ble-languages='c++,f77'

What is the "-lm", and is there anything I can do about it?  I would like to
test this, so I can verify whether or not g77 is working correctly.  I want
to experiment with mixed language programming using fortran and C++ (because
most of my development is based primarily in C++, but there is an incredible
wealth of number crunching libraries available in fortran).  However, if I
encounter a problem, I do not want to have to worry that something I did
when building g77 might have broken it in a way that precludes what I am
attempting from working correctly.

Thanks again,

Ted

-----Original Message-----
From: Mumit Khan [ mailto:khan@NanoTech.Wisc.EDU ]
Sent: Tuesday, July 17, 2001 6:24 PM
To: ted byers
Cc: cygwin@cygwin.com
Subject: Re: building gcc/g++ seems OK but building g77 seems to be
seriously broken


On Tue, 17 Jul 2001, ted byers wrote:

> Well, I have now built GCC3 a couple times, trying to figure out what has
> gone awry, but nothing seems to fix g77 (the summary given below show why
I
> think it is seriously broken).  gcc and g++ seem to be OK since their
> respective test suites give the expected results (the number of unexpected
> failures being extemely small).

When you report gcc build problems, you really need to provide the flags
you passed to configure. Too many variables otherwise.

> 		=== g77 Summary ===
>
> # of expected passes		281
> # of unexpected failures	334
> # of untested testcases		326
> /home/Bted/Objects/gcc/g77 g77 version 3.0 (Fortran Frontend version
0.5.26
> 20010617 (experimental))

Probably a problem in the g77 tests, which passes the -lm, and that causes
a problem on Cygwin 1.1.x.

Regards,
Mumit




--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

Re: building gcc/g++ seems OK but building g77 seems to be seriously broken

[Mumit Khan]

On Tue, 17 Jul 2001, ted byers wrote:

> Well, I have now built GCC3 a couple times, trying to figure out what has
> gone awry, but nothing seems to fix g77 (the summary given below show why I
> think it is seriously broken).  gcc and g++ seem to be OK since their
> respective test suites give the expected results (the number of unexpected
> failures being extemely small).

When you report gcc build problems, you really need to provide the flags
you passed to configure. Too many variables otherwise.

> 		=== g77 Summary ===
> 
> # of expected passes		281
> # of unexpected failures	334
> # of untested testcases		326
> /home/Bted/Objects/gcc/g77 g77 version 3.0 (Fortran Frontend version 0.5.26
> 20010617 (experimental))

Probably a problem in the g77 tests, which passes the -lm, and that causes
a problem on Cygwin 1.1.x.

Regards,
Mumit



--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/

building gcc/g++ seems OK but building g77 seems to be seriously broken

[ted byers]

Well, I have now built GCC3 a couple times, trying to figure out what has
gone awry, but nothing seems to fix g77 (the summary given below show why I
think it is seriously broken).  gcc and g++ seem to be OK since their
respective test suites give the expected results (the number of unexpected
failures being extemely small).

I created logs of the build and tes by redirecting all output into make.log
and test.log.

The only thing I can see of potential concern related to the build of g77 is
the following near the end of stage 2:

mv cp/*.o stage2/cp
if [ -f lang-f77 ]; then \
  mv -f f/*.o f/fini.exe f/stamp-str f/str-*.h f/str-*.j lang-f77 g77spec.o
g77version.o stage2/f; \
fimv: cannot stat `g77spec.o': No such file or directory
mv: cannot stat `g77version.o': No such file or directory

This is puzzling since I can see a number of places earlier in the log where
these two files are compiled.

And the, in test.log, near the beginning of each suite of tests I see the
following warning (and usually the second warning):
WARNING: Couldn't find the global config file.
WARNING: Couldn't find tool init file

I do not know if this means that my installation of cygwin is broken or if
there is something missing from the source code distribution I have for
gcc3.

The following is excerpted from the g77 tests.  Of course, I am not about to
waste bandwidth by sending the entire results, but what appears to be the
salient portions are included here.

(rootme=`pwd`; export rootme; \
srcdir=`cd ../../Source/gcc-3.0/gcc; pwd` ; export srcdir ; \
cd testsuite; \
EXPECT=expect ; export EXPECT ; \
if [ -f ${rootme}/../expect/expect ] ; then  \
   TCL_LIBRARY=`cd .. ; cd ../../Source/gcc-3.0/gcc/../tcl/library ; pwd` ;
\
   export TCL_LIBRARY ; fi ; \
runtest --tool g77 )
WARNING: Couldn't find the global config file.
Test Run By Bted on Sun Jul 15 22:32:58 2001
Native configuration is i686-pc-cygwin

		=== g77 tests ===

Schedule of variations:
    unix

Running target unix
Using /usr/share/dejagnu/baseboards/unix.exp as board description file for
target.
Using /usr/share/dejagnu/config/unix.exp as generic interface file for
target.
Using /home/Bted/Source/gcc-3.0/gcc/testsuite/config/default.exp as
tool-and-target-specific interface file.
Running /home/Bted/Source/gcc-3.0/gcc/testsuite/g77.dg/dg.exp ...
FAIL: g77.dg/20010216-1.f (test for excess errors)
WARNING: g77.dg/20010216-1.f compilation failed to produce executable
FAIL: g77.dg/20010216-1.f (test for excess errors)
WARNING: g77.dg/20010216-1.f compilation failed to produce executable
... lots o' lines - all indicating some kind of failure

		=== g77 Summary ===

# of expected passes		281
# of unexpected failures	334
# of untested testcases		326
/home/Bted/Objects/gcc/g77 g77 version 3.0 (Fortran Frontend version 0.5.26
20010617 (experimental))

make[1]: [check-g77] Error 1 (ignored)

Clearly something is broken in g77 in order for there to be well over two
thirds of the tests either unexpectedly failing or not being run for some
reason.  Is this something I can fix, and if so how?

I have already tried deleting and rebuilding libf2c, but to no effect.

Any suggestions would be welcome.

Thanks,

Ted

R.E. Byers
ted.byers@sympatico.ca


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/