longjmp

Barry Rowlingson B.Rowlingson@lancaster.ac.uk
Mon, 04 Nov 2002 13:47:58 +0000


Warnes, Gregory R wrote:

> Would you mind providing the necessary patch for this behavior?

  Here's the patch, in 'how-to-do-it' format:

1. Get, compile, check, install R as usual.

2. Edit $RSRC/src/include/Defn.h to make sure that the following are 
defined:

# define JMP_BUF jmp_buf
# define SETJMP(x) setjmp(x)
# define LONGJMP(x,i) longjmp(x,i)

  and make sure that the 'sig-' versions are undefined. There's probably 
a way of doing it via configure, but its easy enough to tweak this way.

3. Compile, check. Copy generated binary to $RHOME/bin/R-jmp.bin

4. Edit one line in $RHOME/bin/BATCH so it calls R-jmp.bin instead of 
R.bin.

  Voila, one Mosix-enhanced binary that will only react to the first 
Ctrl-C you throw at it, ignoring subsequent ones. However this isnt 
generally a problem when running with R BATCH since its a batch job and 
there's nobody around to throw Ctrl-Cs at it. However, other signal 
handling may be broken, so dont try to control your R BATCH job by 
sending signals to it.

  Only use this if you know what you are doing, and dont expect any help 
from me or any of the R developers if this breaks stuff - check any 
putative bugs with a standard-compiled version first.

Barry Rowlingson
Maths and Stats
Lancaster University, UK


-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._