Good evening Satish,
Thank you for a Saturday evening reply!
On Sat, 2008-04-05 at 18:36 -0500, Satish Balay wrote:
> Adam,
>
> conffix.h is generated temporarily for this test [based on previous
> tests in configure]. After succesful completion of configure - these
> workarrounds are dumped into bmake/$PETSC_ARCH/petscfix.h You can run
> configure with --doCleanup=0. This option is supporsed to prevent
> deleting temporary files.
Okay, that's what I figured.
> Also there should be an entry in configure.log for a test for mode_t.
> I have the following on my linux box. [if this test fails - configure
> adds an entery for mode_t to conffix.h]
Ah yes, should have known to check configure.log. There's mode_t and
before it size_t, and both are wrong.
Aha, that in turn is caused by the problematic flag -Wno-long-double.
The configure system adds that to its list of compiler flags earlier on
without error, but in the later attempts at compilation, it causes
errors. Why the discrepancy? I suppose that's the toolchain issue
which has created this problem: gcc is confusing the build system!
So I removed -Wno-long-double from python/BuildSystem/compilerOptions.py
and it configures and builds just fine once again. This is merely a
workaround patch for 2.3.3, but it will work for Debian for now.
And now the reply from Matt -- two upstream replies in one Saturday
night! Yes, your diagnosis appears correct, Matt.
Thank you both very much for the quick work, above and beyond the call
of duty! I'll upload a version which closes this and also 474249 later
tonight.
-Adam
> On Sat, 5 Apr 2008, Adam C Powell IV wrote:
>
> > Greetings,
> >
> > Due to a change somewhere in the toolchain, PETSc configuration suddenly
> > stopped working on Debian unstable. The error message in configure.log
> > is:
> >
> > TEST checkSizeof from
> > config.types(/home/hazelsct/petsc-2.3.3/python/BuildSystem/config/types.py:246)
> > TESTING: checkSizeof from
> > config.types(/home/hazelsct/petsc-2.3.3/python/BuildSystem/config/types.py:246)
> > Determines the size of type "typeName", and defines SIZEOF_"typeName" to
> > be the size
> > Checking for size of type: char
> > Pushing language C
> > sh: mpicc -c -o conftest.o -fPIC -Wall -Wwrite-strings -Wno-long-double
> > -Wno-strict-aliasing -g3 conftest.c
> > Executing: mpicc -c -o conftest.o -fPIC -Wall -Wwrite-strings
> > -Wno-long-double -Wno-strict-aliasing -g3 conftest.c
> > sh:
> > Possible ERROR while running compiler: ret = 256
> > error message = {In file included from conftest.c:4:
> > /usr/include/sys/types.h:72: error: conflicting types for ‘mode_t’
> > conffix.h:6: error: previous declaration of ‘mode_t’ was here
> > /usr/include/sys/types.h:88: error: conflicting types for ‘off_t’
> > conffix.h:7: error: previous declaration of ‘off_t’ was here
> > In file included from /usr/include/sys/types.h:147,
> > from conftest.c:4:
> > /usr/lib/gcc/x86_64-linux-gnu/4.2.3/include/stddef.h:214: error:
> > conflicting types for ‘size_t’
> > conffix.h:4: error: previous declaration of ‘size_t’ was here
> > cc1: error: unrecognized command line option "-Wno-long-double"
> > }
> > Source:
> > #include "confdefs.h"
> > #include "conffix.h"
> >
> > #include <sys/types.h>
> > #if STDC_HEADERS
> > #include <stdlib.h>
> > #include <stdio.h>
> > #include <stddef.h>
> > #endif
> >
> > int main() {
> > FILE *f = fopen("conftestval", "w");
> >
> > if (!f) exit(1);
> > fprintf(f, "%lu\n", (unsigned long)sizeof(char));
> > ;
> > return 0;
> > }
> > Compile failed inside link
> >
> > *********************************************************************************
> > UNABLE to CONFIGURE with GIVEN OPTIONS (see configure.log for
> > details):
> > ---------------------------------------------------------------------------------------
> > Unable to determine sizeof_char
> > *********************************************************************************
> >
> > Trouble is, I can't find conffix.h anywhere in order to change it. And
> > "grep conffix petsc-2.3.3/python/BuildSystem/*" and */* etc. doesn't
> > turn up anything useful.
> >
> > How can I investigate further, to try to fix this?
> >
> > -Adam
> >
-Adam
--
GPG fingerprint: D54D 1AEE B11C CE9B A02B C5DD 526F 01E8 564E E4B6
Engineering consulting with open source tools
http://www.opennovation.com/
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
