[R] rw1062

David Khabie-Zeitoune dave at evocapital.com
Wed Jun 25 19:32:34 CEST 2003


On the subject of the LAPACK error -- I also had this problem with the R
(D)COM package. It appears that there is a problem accessing the newer
LAPACK routines rather than the original LINPACK routines which were
built into R. I sent a query to the r-com mailing list and found the
following helpful reply from Roger Bivand fixed my problem (see message
below for the full story...). 

> For the time being, do: solve(..., LINPACK=TRUE)

> This uses the older, built-in functions rather than lapack.dll. I
believe 
> a patch has been made to the R windows release, but we'll have to wait

> until 1.7.1, perhaps, for the path to work correctly.

I have not installed 1.7.1 yet so cannot confirm if this is fixed in
this version, although from what you say it seems not.

-----Original Message-----
From: Roger Bivand [mailto:Roger.Bivand at nhh.no] 
Sent: 14 May 2003 17:19
To: David S. Khabie-Zeitoune
Cc: rcom-l at mailman.csd.univie.ac.at
Subject: Re: [Rcom-l] R Excel complains about not finding lapack.dll


On Wed, 14 May 2003, David S. Khabie-Zeitoune wrote:

> I have just installed the new version of Thomas Baier's RDCOM Server
> package (version 1.2) and the associated Excel interface by Erich 
> Neuwirth (version 1.0). I am using Windows XP Pro.
>  
> I seem to have problems with functions which access the lapack.dll
> such as solve. For example, if I sent a matrix x to R, and I try and 
> run the command "y = solve(x)" from the a cell in Excel using the 
> Excel interface, then I receive an error message:
>  
> >    Error when evaluating
> >    y = solve(x)
> >    evaluation stopped because of an error
>  
> >    Warning: unable to load shared library "C:\Program
> Files\R\rw1070/modules/lapack.dll"
> >        LoadLibrary failure: The specified module cannot be found.
>  
> However, the lapack dll definitely does exist at the specified
> location. Note however the mixture of "back" slashes and "forward" 
> slashes in the path in the error message -- could this be the problem?
>  
For the time being, do: solve(..., LINPACK=TRUE)

This uses the older, built-in functions rather than lapack.dll. I
believe 
a patch has been made to the R windows release, but we'll have to wait 
until 1.7.1, perhaps, for the path to work correctly.

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Breiviksveien 40, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 93 93
e-mail: Roger.Bivand at nhh.no

-----Original Message-----
From: Prof Brian Ripley [mailto:ripley at stats.ox.ac.uk] 
Sent: 25 June 2003 17:02
To: Søren Merser
Cc: R - help
Subject: Re: [R] rw1062


The problem is that you have not put the rw1071/bin directory in your 
path, and so your DCOM-using application is unable to find the dependent

DLLs of R.  This is a problem with the R (D)COM instructions.

On Wed, 25 Jun 2003, Søren Merser wrote:

> I need 'rw1062.zip' since i can't get excel, R1070 or R1071 to work 
> with the R (D)COM Server everything worked fine in the god rw1062 
> days.

Actually, it didn't without the path set.

> the lapack routines can't be loaded
> 'unable to load shared library c:\r/moduleslapack.dll, the specified 
> library could not be found' don't know if it's the forward slashes 
> that anoys windows

They are perfectly valid in Windows, but that `quote' is not the error 
message I get.

> btw, the libraries are in the correct place
> any clues to solve my problem or where to get hold on the zip file??

There never was such a zip file.

-- 
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

______________________________________________
R-help at stat.math.ethz.ch mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-help




More information about the R-help mailing list