[R] How to debug the C programs associated with R-packages?

[Martin Morgan]

On 02/08/2013 05:13 AM, Pulkit Mehrotra wrote:
> hello,
>
> I was having some problems in debugging C codes assocaited with R-packages.
> I don't have much experience with debugging tools used in C language. Is
> there any methods similar to R-debugging tools such as browser  ,debug etc
> which can be used to debug such C files.
>
> I was using the foreign package and found a bug in reading dta files
> containing empty strings "" in the data. I was able to find where the error
> originated in the R file but was unable to do so for the C files.
>
> Any help would be deeply appreciated.
>
> Thank you for your time and consideration.

For debugging package code at the C level, one usually starts (on non-Windows) 
by installing the package without any compiler optimizations, e.g., by following

   RShowDoc("R-admin")

section 6.3.3. setting for instance

   CFLAGS="-g -O0"

in .R/Makevars.

Then when the bug is isolated and easily produced in a script buggy.R start R 
with a C-level debugger such as gdb (this is an old-school command-line style 
debugger, not to everyone's taste)

   R -d gdb -f buggy.R

You'll end up at the gdb prompt

   gdb>

and a typical operation is to (r)un or (c)ontinue execution

   gdb> r

to run buggy.R. You'll end up back in C when there is a segfault, or you press 
cntrl-C, or when you've inserted a breakpoint at some C-level function that you 
suspect is buggy, e.g.,

   > ^C
   gdb> b some_buggy_fun
   gdb> c

When you do end up back in the debugger, you can print C variables or the C 
representation of R variables

   gdb> p c_var
   gdb> call Rf_PrintValue(some_R_variable)

go (u)p and (d)own the call stack, etc. See

   gdb> help

and of course our mutual friend Google.

But this is not for the faint of heart and requires some C-level familiarity. It 
might also be possible to spot the bug with

   R -d valgrind -f buggy.R

which runs quite slowly and will also require C familiarity to interpret output. 
Again it is helpful to run the buggy code with a package that has been installed 
without compiler optimizations. See section 4.3 of RShowDoc("R-exts")

Good luck!

Martin

>
> with regards,
> Pulkit Mehrotra
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> 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.
>


-- 
Computational Biology / Fred Hutchinson Cancer Research Center
1100 Fairview Ave. N.
PO Box 19024 Seattle, WA 98109

Location: Arnold Building M1 B861
Phone: (206) 667-2793