[R] Rwinst.exe problems

[Prof Brian Ripley]

On Mon, 18 Dec 2000, Marc Feldesman wrote:

> I've heard from several users who report no difficulties using the 
> Rwinst.exe installer for the released version of 1.2.0.  I've blown away 
> all traces of 1.2.0 on my system and have re-downloaded all 10 zip files 
> and the installer on both my NT and Win2K boxes.  The memory-overwrite 
> error persists on both systems, suggesting that there is some 
> incompatibility between the 1.2.0 installer (and, BTW, the 1.2.1 and 1.3.0 
> installers as well, but *not* the 1.1.1 version).  The error specifically 

Those pre-release installers are identical to the rw1020 one.

> occurs on all versions whenever I click on the radio button labelled "an 
> add-on package", but not on the button to install the base software.
> 
> I'm trying to isolate software commonalities between the two machines that 
> might explain this problem.  I've cc'd Brian Ripley on these messages in 
> hopes that I might ascertain whether the rwinst.exe installer makes use of 
> any system dependent .dlls that might have been overwritten during recent 
> upgrades to the latest MSIE security patches, Microsoft Office Service 
> Releases, or Norton Antivirus engine and virus signatures.  These are the 
> most plausible culprits.

rwinst.exe uses most of the standard system dlls, plus msvcrt.dll.
pedump will show you exactly. But I suspect the problem is more
prosaic, in that you have an exceptionally long (or in some other way 
unusual) PATH.

The only differences between rwinst for rw1011 and rw1020 are

1) It is compiled against msvcrt.dll.
2) When that radio button is clicked for the first time it

i) Checks if R_HOME is set
ii) if not, looks for ...\rw1020\bin in the PATH

Now, this works for me on several machines and on several OSs (98/NT/W2K)
with and without R_HOME set and/or ...\rw1020\bin in the path.  Try
setting R_HOME and see if this cures it.  If so, my guess is that that the
PATH is malformed (for example, it might exceed the supposed maximum path
length, which my experiments just now suggest is possible).

If setting R_HOME solves the problem, please try a new version I've put at
http://www.stats.ox.ac.uk/pub/bdr/RWin/rwinst.exe
which differs only in not looking at the PATH at all.

-- 
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 272860 (secr)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._