[mesa-users] Compiling Error

[Bill Paxton]

Hi,

Frank pointed out that we can actually have the current ndiff do
both absolute and relative errors, and I've modified the "ck"
scripts accordingly.   So I retract my too-hasty comments
concerning what ndiff could do.    Hmmm, this somehow
sounds familiar --- haven't I given sermons in the past
about users making premature assertions about what
mesa cannot do?  Looks like I'm guilty of such things too!

Cheers,
Bill




On Jul 15, 2013, at 12:38 PM, Francis Timmes wrote:

> bingo.
> 
> might want to modulate your comment about the need to rewrite ndiff ...
> 
> fxt
> 
> 
> On Jul 15, 2013, at 12:37 PM, Bill Paxton <paxton at kitp.ucsb.edu> wrote:
> 
>> 
>> On Jul 15, 2013, at 12:28 PM, Francis Timmes wrote:
>> 
>>> man ndiff suggests the -relerr and -abserr might do what is needed.
>> 
>> Good suggestion.   I'll make the changes in utils/test.
>> e.g., old ck3 line
>> ../../utils/ndiff/bin/ndiff -quiet -relerr 1.0e-3 tmp.txt test_output
>> new one
>> ../../utils/ndiff/bin/ndiff -quiet -relerr 1.0e-3 -abserr 1.0e-30 tmp.txt test_output
>> 
>> That will at least take care of the "fuzzy-zero" cases like the current one.
>> 
>> -B
>