Skip to content

[Rcpp-devel] Making objects in the C++ "side" persistent

7 messages · F.Tusell, David Ibarra Gómez, Jiqiang Guo +2 more

Original
F.Tusell
# 
I have been reading "Exposing C++ functions and classes with Rcpp
modules" and find the idea interesting. However, I read in Section 5
that objects created
using an external pointer are not persistent from session to session.

For my purposes, I think I could use S4 objects that make C++ copies of
themselves at creation time, accessible via an external pointer. All
methods would then make use of the C++ copies for speed, and at the end
of the session the S4 copies that live in the R side would be saved.

I am thinking of writing code which at the start of a session checks all
(S4) objects in the workspace and recreate the C++ copies. Is this the
way to go? Has someone with a similar problem devised an alternative
solution? If anyone can offer some hints (or better yet, point to some
package that I can use for inspiration) I would  be grateful.

Best,


F.Tusell <fernando.tusell at ehu.es>
David Ibarra Gómez
# 
Hello Fernando,

        For sure there will be better ideas, but sometimes I use serialized protocol buffers (RProtoBuf) for persistence, or even to treat data structures (or "classes") between R, C++, python or even C#. You can store serialized protobuf on the file system or using databases (sql or not).

But for sure there are better ideas.

Kind regards
David Ibarra G?mez






-----Mensaje original-----
De: rcpp-devel-bounces at lists.r-forge.r-project.org [mailto:rcpp-devel-bounces at lists.r-forge.r-project.org] En nombre de F.Tusell
Enviado el: jueves, 20 de marzo de 2014 18:40
Para: rcpp-devel at lists.r-forge.r-project.org
Asunto: [Rcpp-devel] Making objects in the C++ "side" persistent

I have been reading "Exposing C++ functions and classes with Rcpp modules" and find the idea interesting. However, I read in Section 5 that objects created using an external pointer are not persistent from session to session.

For my purposes, I think I could use S4 objects that make C++ copies of themselves at creation time, accessible via an external pointer. All methods would then make use of the C++ copies for speed, and at the end of the session the S4 copies that live in the R side would be saved.

I am thinking of writing code which at the start of a session checks all
(S4) objects in the workspace and recreate the C++ copies. Is this the way to go? Has someone with a similar problem devised an alternative solution? If anyone can offer some hints (or better yet, point to some package that I can use for inspiration) I would  be grateful.

Best,


F.Tusell <fernando.tusell at ehu.es>

_______________________________________________
Rcpp-devel mailing list
Rcpp-devel at lists.r-forge.r-project.org
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel

Disclaimer: http://disclaimer.aqualogy.net/
Dirk Eddelbuettel
#
On 20 March 2014 at 18:39, F.Tusell wrote:
| I have been reading "Exposing C++ functions and classes with Rcpp
| modules" and find the idea interesting. However, I read in Section 5
| that objects created
| using an external pointer are not persistent from session to session.

Yes. The memory behind that pointer is not guaranteed...
 
| For my purposes, I think I could use S4 objects that make C++ copies of
| themselves at creation time, accessible via an external pointer. All
| methods would then make use of the C++ copies for speed, and at the end
| of the session the S4 copies that live in the R side would be saved.
| 
| I am thinking of writing code which at the start of a session checks all
| (S4) objects in the workspace and recreate the C++ copies. Is this the
| way to go? Has someone with a similar problem devised an alternative
| solution? If anyone can offer some hints (or better yet, point to some
| package that I can use for inspiration) I would  be grateful.

The Stan folks also did something about this in an extension / rework of
cxxfunction in inline -- see their package cxxfunplus on CRAN. (Which I
haven't used so I can't help with details -- but it help you or point you the
right way.)

Dirk
Jiqiang Guo
# 
RStan doesn't do anything to persist any c++ object.  What's done is to
save the binary file created using package inline in an R object so it can
be reused without recompiling c++ code.  So what is in cxxfunplus might not
be related.

Jiqiang
On Mar 20, 2014 11:53 AM, "Dirk Eddelbuettel" <edd at debian.org> wrote:

            

      
      
      
    

        

      
      
    

        
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/rcpp-devel/attachments/20140320/b1dc1d9c/attachment.html>

  
  


  


      

    

    

      
      

        


  

        

      Dirk Eddelbuettel
    

    

      
      #
    

  


  

      
On 20 March 2014 at 12:40, Jiqiang Guo wrote:

        
| RStan doesn't do anything to persist any c++ object.? What's done is to save
| the binary file created using package inline in an R object so it can be reused
| without recompiling c++ code.? So what is in cxxfunplus might not be related.

Thanks for the clarification.

My recommendation is usually to just create a package. We know that works.

Dirk

  
    

      
      
    

  


  


      

    

    

      
      

        


  

        

      Romain Francois
    

    

      
      #
    

  


  

      
Creating a package does not give you persistence of external pointers across sessions. 

Le 20 mars 2014 ? 19:52, Dirk Eddelbuettel <edd at debian.org> a ?crit :

            

      
      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Dirk Eddelbuettel
    

    

      
      #
    

  


  

      
On 20 March 2014 at 20:03, Romain Fran?ois wrote:

        
| Creating a package does not give you persistence of external pointers across sessions. 

I didn't claim it did.  AFAICR nothing simple achieves that.

Dirk