[R] R 2.0.0 not suffisantly reliable to be be used

Mike Prager Mike.Prager at noaa.gov
Mon Oct 11 15:44:24 CEST 2004


>On Sun, 10 Oct 2004, Fan wrote:
>
> > For occasional users, I would say, there's no worst thing than that:
> > you installed the new release, and soem of your existing codes no
> > longer work !

There is a close parallel here.  On comp.lang.fortran, a frequent problem 
is that existing code doesn't work with a new compiler. Such complaints are 
often from occasional Fortran users, those who have inherited code from 
others, or those who have been using a single compiler for 20 years and 
have finally updated.

In the vast majority of cases, the code in question is not standard 
conforming.  Either (1) the old compiler ignored various syntax errors, (2) 
it gave the right answer by luck (i.e., the programmer made assumptions 
that the compiler met, but that Fortran in general does not meet), or (3) 
the old release gave the wrong answers, which were never checked.

So, in response to Fan's assertion, I say:  NO!  It's often a good thing 
when existing codes no longer work.  That provides an opportunity to fix 
them. The worst thing is to have bad code run and give the wrong answer for 
the question (you think) you're asking.


-- 
Michael Prager, Ph.D.                <Mike.Prager at noaa.gov>
NOAA Beaufort Laboratory
Beaufort, North Carolina  28516
http://shrimp.ccfhrb.noaa.gov/~mprager/
***




More information about the R-help mailing list