[MITgcm-support] RE: MITgcm

[Chris Hill]

Jane,

 I'm told that

  -xassembler-with-cpp 

 also works to make xlf happy with the slightly broken cpp in Panther.
 

Chris

> -----Original Message-----
> From: Jane Jimeian [mailto:jane at atlantic.ocean.fsu.edu] 
> Sent: Friday, June 25, 2004 4:49 PM
> To: 'Jane Jimeian'; Chris Hill
> Cc: support at mitgcm.org; dewar at ocean.fsu.edu
> Subject: RE: MITgcm
> 
> i'm using the IBM xlf compilers.  i believe cpp is GNU 
> (installed development pkg Xcode).
> i had to fix the line that cpp puts in the beginning of the file:
> 
>         #pragma GCC set_debug_pwd 
> 
> by changing the Makefile to include:
> 
>         # Fix #pragma line
>         SED = | sed -e s/\#pragma/C\#pragma/
>         
>         [snip]
>         
>         .F.f:
>                 $(CPP) $(DEFINES) $(INCLUDES) $(SED) > $@
> 
> i think the line (#pragma...) that cpp puts is new thing with 
> Xcode for Panther (v10.3)
> 
> http://developer.apple.com/documentation/ReleaseNotes/Develope
> rTools/GCC3.html
> 
>         The GCC 3.3 preprocessor inserts a new pragma, 
> #pragma GCC  set_debug_pwd,
>         as part of the new Distributed Builds feature. (See 
> below.) This may surprise
>         tools and scripts that  depended on the exact form of 
> preprocessed output
>         from GCC. These  scripts should be rewritten to 
> ignore unrecognized pragmas.
>  
>  > From: "Chris Hill" <cnh at mit.edu>
>  > To: "'Jane Jimeian'" <jane at atlantic.ocean.fsu.edu>  > Cc: 
> <dewar at ocean.fsu.edu>, <support at mitgcm.org>  > Subject: RE: 
> MITgcm  > Date: Fri, 25 Jun 2004 15:16:39 -0400  >  > Hi 
> Jane,  >  >  Things should build on a G5. I'm forwarding this 
> to our support list (  > support at mitgcm.org ). There are 
> several G5 users on that list whih may  > recognise the 
> error. That's also the place to send support questions first.
>  > That way several people see them.
>  >
>  >  One question - are you using xlf (the IBM compilers) or 
> the GNU compilers (  > g77 etc... ). Both are available for 
> G5, but you need to use a different  > optfile for each one.
>  >
>  > Thanks,
>  >
>  > Chris
>  >   
>  >
>  > > -----Original Message-----
>  > > From: Jane Jimeian [mailto:jane at atlantic.ocean.fsu.edu]
>  > > Sent: Friday, June 25, 2004 3:00 PM  > > To: cnh at mit.edu 
>  > > Cc: dewar at ocean.fsu.edu  > > Subject: MITgcm  > >  > >  
> > > i work for bill dewar.  i'm trying to build MITgcm on a 
> G5  > > (dual processer) and i'm having a problem 
> interpreting the  > > documentation in  > > 
> ~/MITgcm/tools/build_options/darwin_ppc_xlf:
>  > > 
>  > > #   o tim.c should use uembellished cloc() are the procedure 
>  > > name for XLF
>  > > #     ETIME (timers.F) and FDATE (utils.F) are not available 
>  > > with Darwin+XLF so need to be switched  > >  > > this is 
> where it stops in the makefile:
>  > >
>  > > [snip]
>  > > _myfield.o write_state.o tim.o timer_stats.o  > > 
> /usr/bin/ld: Undefined symbols:
>  > > _etime
>  > > make: *** [mitgcmuv] Error 1
>  > > 
>  > > thanks for any help you can offer.   ~jane
>  > >
>  >
>  > 
>