[R] R.dll

[Duncan Murdoch]

On Tue, 03 Aug 2004 08:59:47 +1000, Ashton DeSilva
<Ashton.DeSilva at Buseco.monash.edu.au> wrote:

>Hello
>
>After conducting a simulation R closes down reporting the problem I have
>listed below.  These are extracted from the event viewer option under
>the control panel option in windows.   This problem occurs quite often
>but is as far as I can tell fairly sparadic. I can run identicle
>programs which run successfully the first time but on the second run it
>does not (and vise-versa also).   Furthermore if I was to run this on
>Windows 2000 then the problem doesn't seem to occur at all. 
>
>Each fault happened at separate times:
>Faulting application rgui.exe, version 1.91.30621.0, faulting module
>r.dll, version 1.91.30621.0, fault address 0x00132900.

You need to come up with code that reproduces the error before anyone
else is going to be able to help you with it.  Since you say this
happens during a simulation, the differences between runs may be due
to different random number seeds.  You could try things like

set.seed(1)
# Run simulation
set.seed(2)
# Run simulation

...

until you find a value that triggers the crash, then see if it happens
reproducibly.  If so, try to narrow things down:  print diagnostics so
you know how much of the simulation succeeded before the crash
happened.  Save .Random.seed just before the crash, and see if running
just 

.Random.seed <- saved.value
# Run line that caused the crash

causes the crash reproducibly.  If so, you'll have a nice bug report
to submit (or may have found a bug in your own code, or some
contributed package).

Duncan Murdoch