[MITgcm-devel] solasrv4 testing

[Jean-Michel Campin]

Martin,

No problem. OK like this.

Jean-Michel

On Mon, Nov 08, 2010 at 02:28:52PM +0100, Martin Losch wrote:
> Hi Jean-Michel,
> 
> I'd rather not set CC explicitly because the path keeps changing in an uncontrolled way when the compiler suite is updated on that computer. Hope that that's OK with you.
> 
> Martin
> 
> On Nov 8, 2010, at 2:14 PM, Jean-Michel Campin wrote:
> 
> > Hi Martin,
> > 
> > It looks OK to me, and could leave things as they are for now
> > (may be at some point it could be good to also set CC when
> > we specify CFLAGS in the optfile).
> > Thanks,
> > Jean-Michel
> > 
> > On Mon, Nov 08, 2010 at 11:58:36AM +0100, Martin Losch wrote:
> >> This seems to be solved, right? Should I do anything like change the opt file for solasrv?
> >> 
> >> M.
> >> 
> >> On Nov 5, 2010, at 2:32 PM, Jean-Michel Campin wrote:
> >> 
> >>> Martin,
> >>> 
> >>> I think I understand the problem:
> >>> The optfile "sunos_i86pc_f95" specify CFLAGS but not CC,
> >>> and "look_for_C_compilers" does not use CFLAGS to test
> >>> the possible compiler, so that it can select gcc even
> >>> if it does not work with CFLAGS.
> >>> 
> >>> Solution:
> >>> a) A quick fix: I can change the order of CC candidate in the
> >>> look_for_C_compilers
> >>> b) add CFLAGS when testing the CC candidate. Going to check if
> >>> is compatible with other use of CC in genmake2.
> >>> 
> >>> Cheers,
> >>> Jean-Michel
> >>> 
> >>> On Fri, Nov 05, 2010 at 09:11:55AM -0400, Jean-Michel Campin wrote:
> >>>> Hi Martin,
> >>>> 
> >>>> I have the impression that my latest changes in genmake2 could
> >>>> have something to do with the recent fail test on solasrv4
> >>>> (does not compile C subroutines). Can you confirm ?
> >>>> 
> >>>> It was clear to me that my changes would not change anything
> >>>> when the C compiler is set in the optfile (CC=...). 
> >>>> In the case of solasrv4, CC is not set in the opfile, but I thought
> >>>> it would try to find a good one. I am interested in knowing
> >>>> what is wrong in the "look_for_C_compilers" part of genmake2
> >>>> (I am going to change testreport to return genmake.log +
> >>>> genmake2.tr_log in case the built fail).
> >>>> The part "look_for_C_compilers" was in previous genmake2 
> >>>> but only used when no optfile was specified.
> >>>> 
> >>>> Cheers,
> >>>> Jean-Michel
> >>>> 
> >>>> 
> >>>> 
> >>>> _______________________________________________
> >>>> MITgcm-devel mailing list
> >>>> MITgcm-devel at mitgcm.org
> >>>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
> >>> 
> >>> _______________________________________________
> >>> MITgcm-devel mailing list
> >>> MITgcm-devel at mitgcm.org
> >>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
> >> 
> >> 
> >> _______________________________________________
> >> MITgcm-devel mailing list
> >> MITgcm-devel at mitgcm.org
> >> http://mitgcm.org/mailman/listinfo/mitgcm-devel
> > 
> > _______________________________________________
> > MITgcm-devel mailing list
> > MITgcm-devel at mitgcm.org
> > http://mitgcm.org/mailman/listinfo/mitgcm-devel
> 
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel