Re: [Angelo.Graziosi: Problems with patched gnu make ? (was:Need Volunteers to test patch for gnu make)]

[Angelo Graziosi]

The attached file contains the logs requested.


Cheers,
   
   Angelo.

On Tue, 3 Oct 2006, Eli Zaretskii wrote:

> > Date: Tue, 3 Oct 2006 00:12:08 +0200 (MET DST)
> > From: Angelo Graziosi <address@hidden>
> > cc: address@hidden
> > 
> > The attached file contains the new logs obatined as follow:
> > -------------------------------------------------------------
> > $ ls -lrt /bin/make.*
> > ...           151552 Jul 10 00:08 make.orig.exe
> > ...           494678 Sep  8 21:28 make.patched.exe
> > ...               16 Oct  2 22:43 make.exe -> make.patched.exe
> > 
> > cd ~/Downloads/GFortran/gcc/.build
> > ../configure  ....
> > make
> > 
> > 
> > IT FAILS as described...
> > 
> > cd /home/Angelo/Downloads/GFortran/gcc/.build/gcc
> > make -p --debug=i 2>&1 | tee build-gfortran-make-patched.log
> > 
> > *** Changing MAKE *** 
> > 
> > $ ls -lrt /bin/make.*
> > ...          151552 Jul 10 00:08 make.orig.exe
> > ...          494678 Sep  8 21:28 make.patched.exe
> > ...              13 Oct  2 23:49 make.exe -> make.orig.exe
> > 
> > 
> > 
> > make -p --debug=i 2>&1 | tee build-gfortran-make-unpatched.log
> > -------------------------------------------------------------
> 
> Thanks.
> 
> It looks like my guess about vpath was correct: the failing Make
> doesn't see it.  That's quite strange, since I cannot see anything in
> the code that could explain the problem.  Obviously, I'm missing
> something...
> 
> But before I dig deeper into the code that processes vpath, I'd like
> to be absolutely sure I'm looking in the right place.  So I'd like to
> ask you to run one more test.  At the end of this message you will
> find a short Makefile, that is a very small subset of the Makefile you
> have in the /home/Angelo/Downloads/GFortran/gcc/.build/gcc directory
> in your gfortran source tree.  Please save the file below as
> /home/Angelo/Downloads/GFortran/gcc/.build/gcc/cygdos-test.mk, chdir
> to /home/Angelo/Downloads/GFortran/gcc/.build/gcc, and run the
> following command:
> 
>     make -p --debug=i -f cygdos-test.mk
> 
> with both the patched and unpatched Make.  If I'm right, the patched
> Make should fail with the same or similar error message as you saw in
> your build attempt: it should complain about "no rule to make
> `cpp.texi', needed by `doc/cpp.info'"; whereas the unpatched Make
> should succeed and run `makeinfo'.
> 
> Whatever the results are, please send me the two logs produced by the
> above command with each of the two versions of Make.  Thanks.
> 
> ------------------------- cut here ---------------------------------------
> vpath %.texi ../../gcc/doc:../../gcc/doc/include
> 
> doc/cpp.info: cpp.texi
> 
> doc/%.info: %.texi
>       makeinfo -I . -I ../../gcc/doc -I ../../gcc/doc/include -o $@ $<
> --------------------------------------------------------------------------
>

cygdos-test.tar.bz2
Description: Binary data