Skip to content

Converting libraries for different operating systems

4 messages · lamorim at email.unc.edu, Dirk Eddelbuettel, Michael Rutter +1 more

Original
lamorim at email.unc.edu
# 
Dear list members,

    I used a R package some years ago with an older version of R in 
sparc-sun-solaris2 (Unix) machine. I have to redo some of those 
analysis, but now my machine is ubuntu. I'm getting the following error 
message when I try to install the library:

Error: package crf was built for sparc-sun-solaris2.6

    Is there any way to convert the library for different operating 
system (Linux in this case)? If so, could you please give some 
directions?

                      Thank you,

                             Leila
Dirk Eddelbuettel
#
On 22 September 2009 at 07:13, lamorim at email.unc.edu wrote:
| Dear list members,
| 
|     I used a R package some years ago with an older version of R in 
| sparc-sun-solaris2 (Unix) machine. I have to redo some of those 
| analysis, but now my machine is ubuntu. I'm getting the following error 
| message when I try to install the library:
| 
| Error: package crf was built for sparc-sun-solaris2.6
| 
|     Is there any way to convert the library for different operating 
| system (Linux in this case)? 

Yes, it is called "install from source".  

Seriously, that is what a compiler does: translate from (human-readable)
source code to (machine-readable) executables.  And no, these are generally
not portable across system.  

| If so, could you please give some directions?

You need

    sudo R CMD INSTALL foo_1.2-3.tar.gz

and you probably want to install the r-base-dev package first, as well as
everything you need for building foo.

Dirk
Michael Rutter
# 
There may be a hitch.  I cannot find any evidence of a package called 
'crf'.  If all you have is the binary from your Sparc machine, that will 
not work under Ubuntu.  You are going to need the source for that 
package.  Maybe there is a current package that is similar to crf.

Michael
Dirk Eddelbuettel wrote:
denis brion
# 
+ If you have the sources, it is feasible and likely to be easy :

if there are no *.c , *f files it depends on, but "only" R scripts(*.R"=, it is  easy, 
else, it might be complicated but it is not desesperate case (I sometimes have to write things like that). 

Anyway, one can have private packages one builds oneself (else, it would be impossible to test packages before making them standard)...
Detailed explanations can be found in 
http://cran.r-project.org/doc/manuals/R-exts.html

(and colleagues||friends||the_lazy_web can help you, if it is something you find this direction complicated)

+ If you have "only" binaries, I do not see any solution except reminding what it did, and finding out/rewriting  the same functionalities. 

Good luck and have nice days
Denis

--- En date de?: Mar 22.9.09, lamorim at email.unc.edu <lamorim at email.unc.edu> a ?crit?: