[R] R_alloc with structures with "flexible array members"

[Ramon Diaz-Uriarte]

Humm... but I mistakenly thought I was assuming almost nothing.

R.



On Wed, Mar 5, 2008 at 10:38 PM, Jeffrey Horner
<jeff.horner at vanderbilt.edu> wrote:
> Ramon Diaz-Uriarte wrote on 03/05/2008 03:00 PM:
>
> > Dear Prof. Ripley,
>  >
>  > Yes, of course! You are right. What a silly mistake on my part! I was
>  > using a standalone program for development of functions, debugging,
>  > etc, of what is part of a package.
>
>  Aha! The lesson I take away from this then is question everything,
>  assume nothing.
>
>  Jeff
>
>
>
>  >
>  > Thanks,
>  >
>  > R.
>  >
>  >
>  >
>  > On Wed, Mar 5, 2008 at 8:45 PM, Prof Brian Ripley <ripley at stats.ox.ac.uk> wrote:
>  >> On Wed, 5 Mar 2008, Ramon Diaz-Uriarte wrote:
>  >>
>  >>  > Dear Jeff,
>  >>  >
>  >>  > Thanks for the suggestion. However, something is still not working.
>  >>  > This is a simple example:
>  >>  >
>  >>  > ***************************  start C ************
>  >>  > #include <R.h>
>  >>  >
>  >>  > struct Sequence {
>  >>  >  int len;
>  >>  >  unsigned int state_count[];
>  >>  > };
>  >>  >
>  >>  >
>  >>  > int main(void) {
>  >>  >
>  >>  >  struct Sequence *A;
>  >>  >  int n = 4;
>  >>  >
>  >>  >  // First line segfaults. Second doesn't
>  >>  >    A = (struct Sequence *) R_alloc(1,  sizeof(struct Sequence) + n *
>  >>  > sizeof(unsigned int));
>  >>  >  //  A = malloc(sizeof(struct Sequence) + n * sizeof(unsigned int));
>  >>  >
>  >>  >  return(0);
>  >>  > }
>  >>  >
>  >>  > ***********  end C  **********
>  >>  >
>  >>  >
>  >>  > I then do
>  >>  > gcc -std=gnu99 -Wall -I/usr/share/R/include -I/usr/share/R/include
>  >>  > -L/usr/lib/R/lib -lR ex7.c
>  >>  >
>  >>  > and the ./a.out segfaults when I use R_alloc (not with malloc).
>  >>
>  >>  You can't use R_alloc in a standalone program without initializing R,
>  >>  which has not been done here.
>  >>
>  >>  You said 'in a package', but this is not in a package.
>  >>
>  >>
>  >>
>  >>  >
>  >>  >
>  >>  > Best,
>  >>  >
>  >>  > R.
>  >>  >
>  >>  > On Wed, Mar 5, 2008 at 5:23 PM, Jeffrey Horner
>  >>  > <jeff.horner at vanderbilt.edu> wrote:
>  >>  >> Ramon Diaz-Uriarte wrote on 03/05/2008 04:25 AM:
>  >>  >>
>  >>  >>
>  >>  >>> Dear All,
>  >>  >> >
>  >>  >> > In a package, I want to use some C code where I am using a structure
>  >>  >> > (as the basic element of a linked list) with flexible array members.
>  >>  >> > Basically, this is a structure where the last component is an
>  >>  >> > incomplete array type  (e.g., Harbison & Steel, "C, a reference
>  >>  >> > manual, 5th ed.", p. 159) such as:
>  >>  >> >
>  >>  >> > struct Sequence {
>  >>  >> >   struct Sequence *next;
>  >>  >> >   int len;
>  >>  >> >   unsigned int state_count[];
>  >>  >> > };
>  >>  >> >
>  >>  >> >
>  >>  >> > To create one such sequence, I allocate storage (following Harbison
>  >>  >> > and Steel) in a C program as follows:
>  >>  >> >
>  >>  >> > struct Sequence *A;
>  >>  >> > int n = 4;
>  >>  >> > A = malloc( sizeof(struct Sequence) + n * sizeof(unsigned int));
>  >>  >> >
>  >>  >> >
>  >>  >> > If I understand correctly, however, it would be better to use R_alloc
>  >>  >> > instead of malloc (memory automagically freed on exit and error;
>  >>  >> > error-checking). But I do not know how to make the call to R_alloc
>  >>  >> > here, since R_alloc allocates n units of size bytes each.
>  >>  >> >
>  >>  >> >
>  >>  >> > I've tried, without success, the following two:
>  >>  >> >
>  >>  >> > int to_add_for_R_alloc =
>  >>  >> >     (int) ceil((float) sizeof(struct sequence) / sizeof(unsigned int));
>  >>  >> >
>  >>  >> >   A = (struct sequence *) R_alloc(to_add_for_R_alloc + n,
>  >>  >> > sizeof(unsigned int));
>  >>  >> >
>  >>  >> > or even a brute force attempt as:
>  >>  >> >
>  >>  >> >  A = (struct sequence *) R_alloc( 100,  sizeof(struct sequence));
>  >>  >> >
>  >>  >> >
>  >>  >> > but both result in segmentation faults.
>  >>  >> >
>  >>  >> >
>  >>  >> > Should I just keep using malloc (and free at end)?
>  >>  >>
>  >>  >>  Hi Ramon,
>  >>  >>
>  >>  >>  You should be able to use R_alloc without seg faults, so there's
>  >>  >>  something wrong with your code somewhere. R_alloc multiplies its
>  >>  >>  arguments together to come up with the total number of bytes to allocate
>  >>  >>  then it allocates a raw vector and returns the data portion.
>  >>  >>
>  >>  >>  So you can just treat R_alloc similarly to malloc by calling
>  >>  >>  R_alloc(1,sizeof(struct Sequence) + n * sizeof(unsigned int)).
>  >>  >>
>  >>  >>  Best,
>  >>  >>
>  >>  >>  Jeff
>  >>  >>  --
>  >>  >>  http://biostat.mc.vanderbilt.edu/JeffreyHorner
>  >>  >>
>  >>  >
>  >>  >
>  >>  >
>  >>  > --
>  >>  > Ramon Diaz-Uriarte
>  >>  > Statistical Computing Team
>  >>  > Structural Biology and Biocomputing Programme
>  >>  > Spanish National Cancer Centre (CNIO)
>  >>  > http://ligarto.org/rdiaz
>  >>  >
>  >>  > ______________________________________________
>  >>  > R-help at r-project.org mailing list
>  >>  > https://stat.ethz.ch/mailman/listinfo/r-help
>  >>  > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
>  >>  > and provide commented, minimal, self-contained, reproducible code.
>  >>  >
>  >>
>  >>  --
>  >>  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
>



-- 
Ramon Diaz-Uriarte
Statistical Computing Team
Structural Biology and Biocomputing Programme
Spanish National Cancer Centre (CNIO)
http://ligarto.org/rdiaz