[RkWard-devel] Another preview, last call for testing

[Thomas Friedrichsmeier]

> Okay Thomas, I added the line you specified, recompiled and
> reinstalled.  I'm testing the crash by running library(Biobase) -
> this then loads the required package "tools".  This time I loaded
> "tools" first, with no crashes.  Then I loaded Biobase and it crashed
> in the typical manner.

Ok, thanks for your testing. You can undo the last two changes, apparently 
this is not the place look for the cause of the crash.

> I don't recall whether I've ever successfully run rkward (v. 0.3.2,
> etc.) without this bug on FreeBSD (I've run rkward on a lot of
> different platforms and sometimes my memory gets fuzzy).
>
> Please let me know if I can do anything else.

Admittedly, I'm a bit short on ideas on this subject. However, this leaves my 
gut-feeling, it is somehow related to the problem with locating libR.so (even 
though a wrong setting for R.home () does not seem to be the cause of that). 
On the RPy-pages I found some tips about libR.so, that might be related:

"During the compilation of RPy, the path to libR.so is included in the RPy 
shared library. This should allow RPy to find libR.so unless it is moved to a 
new location, in which case you will need to do one of the following 
(substitute R_HOME for the directory where R is installed, 
usually /usr/local/lib/R): 

1) make a link to R_HOME/bin/libR.so in /usr/local/lib, or in some other 
location that your system knows as a standard place for libraries; then run 
'ldconfig' as root; 

2) or edit /etc/ld.so.conf to add the line 'R_HOME/bin' and run 'ldconfig' as 
root; 

3) or set the environment variable LD_LIBRARY_PATH with 
    export LD_LIBRARY_PATH=$LD_LIBRARY_PATH\
           :R_HOME/bin

 in your .bashrc (or equivalent). 
 If these suggestions don't work, try to remove the R source directory, unpack 
the sources again and then execute the previous instructions. It has been 
reported that the removing and unpacking may be necessary."

I think you're currently basically using option 1) (although with a copy 
instead of a link). Maybe you could remove that copy again for testing 
purposes, and try your luck with approach 2) or 3).
Before testing it with the Biobase-library, you might want to try loading some 
other libraries, that previously complained about not being able to find 
libR.so (as described in 
http://sourceforge.net/mailarchive/message.php?msg_id=13338572).
I won't promise anything, but I think this should be worth a try.

Thanks!
Thomas