[R] R 1.7.1 arima0 problem

[Prof Brian Ripley]

Note that those examples are run by `make check' during the installation
process, so the installation should have failed.  How come that was not
detected earlier?

RedHat 6.2 is rather old these days and as I recall came with flaky 
compilers (the infamous gcc `2.96').  Was a released version of gcc used: 
if not, please try one.  In any case, this is a sign of a broken R 
installation, and not a breakage that has been reported before.

On Mon, 4 Aug 2003, Uwe Ligges wrote:

> Yu Zhang wrote:
> 
> > Hi, I'm trying to go through the examples for function
> > arima0() in ts package, i.e,
> > 
> >>data(lh)
> >>arima0(lh, order = c(1,0,0))
> > 
> > each time the call to arima0() causes a segmentation
> > fault. I checked the earlier version (1.1.1) of R,
> > the function arima0 works fine. 
> > 
> > Tracing the call indicates that the function
> > "setup_starma" (in pacf.c under ts) interprets 
> > the addresses of the arguments incorrectly.
> > The problem seems to 
> > be related to INTEGER(na), which re-align the address
> > of na which otherwise would be correct. Is it a known
> > problem with a quick fix or system-dependent? FYI, I'm
> > 
> > using R version 1.7.1 on redhat 6.2.
> > 
> > Any hints will be appreciated. 
> 
> Works for me on Windows. There is a function arima these days, did you 
> try that one as well?

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595