[MITgcm-devel] new version of TAF

[Jean-Michel Campin]

Hi Patrick,

On Fri, Jan 30, 2009 at 07:43:28AM -0500, Patrick Heimbach wrote:
>
> Hi there,
>
> the TAF problem has been fixed for new version 1.9.40.
>
> Incidentally (not fully related to this test), one question:
> in testreport, what does "N/O" mean in tr_out.txt line
> Y Y Y Y -->--<N/O

Looks like the run finished normally (the 4th "Y"),
but the number of "cost" & "gradient" output that are found in the 
new output file (could even be zero) is no the same as in the 
reference output file, so that you don't get any comparison number 
and just 2 times "--" instead. 

Jean-Michel

>
> -p.
>
>
>
> On Jan 24, 2009, at 6:44 PM, Jean-Michel Campin wrote:
>
>> Hi Patrick,
>>
>> The new version of TAF (1.9.39, the default since Jan 23) does not
>> work for most of our test case: curently tested on eddy:
>> http://mitgcm.org/testing/results/2009_01/tr_eddy_20090124_0/ 
>> summary.txt
>>
>> With the same code, the previous version (1.9.38) has no problem:
>> currently tested on faulks:
>> http://mitgcm.org/testing/results/2009_01/tr_faulks_20090124_0_0/ 
>> summary.txt
>> (except for bottom_ctrl_5x5 and tutorial_global_oce_optim which were
>>  using the new version, but this should be fixed for tomorrow testing)
>>
>> When you have time to look at this (on may be ask Ralf G. why
>> he changed it).
>>
>> I modified genmake2 for this testing on faulks, and could be
>> a good idea not to check-in changes in genmake2 until this TAF
>> issue is solved.
>>
>> Cheers,
>> Jean-Michel
>>
>> _______________________________________________
>> MITgcm-devel mailing list
>> MITgcm-devel at mitgcm.org
>> http://mitgcm.org/mailman/listinfo/mitgcm-devel
>
> ---
> Patrick Heimbach | heimbach at mit.edu | http://www.mit.edu/~heimbach
> MIT | EAPS 54-1518 | 77 Massachusetts Ave | Cambridge MA 02139 USA
> FON +1-617-253-5259 | FAX +1-617-253-4464 | SKYPE patrick.heimbach
>
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel