[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: reading NaNs



From: Aubrey Jaffer <agj@xxxxxxxxxxxx>
Subject: Re: reading NaNs
Date: Tue, 25 Oct 2005 18:45:29 -0400 (EDT)

> If program A writes out its state, it would be useful to see that NaNs
> were computed.  It gives operators a chance to capture the use case
> which provoked the error.  If the program state is very valuable, then
> it can be repaired manually.
> 
> But if program B reads its initial state from the file, its reading of
> NaNs puts errors into its state which can propagate and corrupt it.

It depends.   Program B may decide to discard the part of the
result of program A which includes NaN and continue to operate,
or replacing NaNs with special values to indicate anormality.
There are cases that program A cannot decide by itself whether its
result including NaN is useful by the downstream or not.  There
are cases that manual repairment of NaNs are too expensive.
I had such experiences in the production environment.

So I would need to read NaNs safely.  However, if having
read-syntax for NaN hurts integrity of the numeric system
as a standard, I can live with srfi-10 (e.g. #,(nan)).

--shiro