Re: Difference between NaN and NA?

[Jaroslav Hajek]

On Fri, Apr 9, 2010 at 8:24 AM, Søren Hauberg <address@hidden> wrote:
> fre, 09 04 2010 kl. 08:07 +0200, skrev Matthias Brennwald:
>> On Apr 9, 2010, at 7:44 AM, Søren Hauberg wrote:
>>
>> > So, I guess my basic question is: is it even possible to create a class
>> > that deals with NaN's (or NA's) entirely behind the users back?
>>
>> I did not fully understand everything in previous discussion of the
>> technical details about what might be changed in Octave and how to do
>> it. However, I have a clear opinion (if not a wish and hope) on the
>> above question: I (the user) would like to keep control over how NaN
>> and NA values are treated. I prefer doing checks on the existence of
>> NaN or NA values in my data, and explicitly treat these values the way
>> I want. No black magic, please!
>
> Just to be clear on the suggestion: it would require the user to
> manually ask for NA-treatment. If we had this class, you would have to
> do something like
>
>  data = load ("mydata.txt");
>  NA_safe_data = NA_friendly_matrix (data);
>  # Process 'NA_safe_data'
>
> So, any potential black magic would require the user to ask for it.
>

Yes. It would just have to be done once, instead of passing a flag to
each function.
I'd say forget about this. I don't mind having the current limited
support of NA's in Octave. Users should just be aware that the
treatment of NA's is inconsistent and very limited, but maybe they
already are.

-- 
RNDr. Jaroslav Hajek, PhD
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz