[R] Re: Windows versus Unix packages in CRAN ...

[Peter Dalgaard]

Duncan Murdoch <dmurdoch at pair.com> writes:

> On Fri, 21 May 2004 15:41:18 +0200, Tamas Papp <tpapp at axelero.hu>
> wrote:
> 
> >You are probably right in saying that they _could_ have done better,
> >but I would not use "should" in this context.  AFAIK the package is
> >free (as in beer) software, which means that you are not paying for
> >it.  The maintainers probably do not need a Linux version (yet), so it
> >was not easy to use it under Linux.  Feel free to contribute.  
> 
> Yes, indeed!  

Well, yes and no. Yes, people should feel free to help out with the
maintenance, but no, it is not reasonable to leave cross-platform
issues unaddressed.

Go look on Statlib and I'm sure you'll find code that built and ran
with HP-UX Fortran vintage 1992 or so, and got submitted with exactly
that sentiment: "It works for me, it's yours for free, but you have to
do the work." Of course, not even the original author would know how
to make it work again by now. This kind of "bitrot" is what CRAN and
all of its red tape was designed to avoid.

Whereas we cannot reasonably require people to test code on platforms
that they haven't got, we can and should require them to adhere to
reasonable standards and test procedures (which, mind you, other
people have invested a serious amount of time in working out).

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907