R devel - Apr 2003 - Rgui crashes after setting par(ask=TRUE) and using lattice (PR#2801)

Full_Name: Scot McNary
Version: 1.7
OS: Windows 98se
Submission from: (NULL) (134.192.131.246)


Hi,

Rgui causes a page fault when asking for lattice graphics after setting
par(ask=TRUE).  This doesn't happen in 1.6.2, but does in 1.7.0.

library(lattice)
par(ask=TRUE)
example(dotplot)

Windows "Details" about crash:

RGUI caused an invalid page fault in
module R.DLL at 017f:00421ce0.
Registers:
EAX=00421cd0 CS=017f EIP=00421ce0 EFLGS=00010206
EBX=017b9e18 SS=0187 ESP=0096d7b4 EBP=0096d7d8
ECX=1177ab43 DS=0187 ESI=0096d8c8 FS=8397
EDX=00000000 ES=0187 EDI=0096d8d0 GS=0000
Bytes at CS:EIP:
8b 9a d0 00 00 00 52 d9 7d fa dc 05 c0 1c 42 00 
Stack dump:
00000003 00000000 0096d820 01be3200 17b70d3f 0c705027 88080012 01387cf4 017b9e18
0096d838 11771699 0096d818 0096d820 0096d828 0096d830 01c1e6e8 

> version
         _              
platform i386-pc-mingw32
arch     i386           
os       mingw32        
system   i386, mingw32  
status                  
major    1              
minor    7.0            
year     2003           
month    04             
day      16             
language R

On Sun, 20 Apr 2003 21:01:41 +0200 (MET DST), you wrote:
>Full_Name: Scot McNary
>Version: 1.7
>OS: Windows 98se
>Submission from: (NULL) (134.192.131.246)
>
>
>Hi,
>
>Rgui causes a page fault when asking for lattice graphics after setting
>par(ask=TRUE).  This doesn't happen in 1.6.2, but does in 1.7.0.
I don't get the crash in XP.  I'll try it tomorrow in Win98.

Duncan Murdoch

On Sun, 20 Apr 2003 21:01:41 +0200 (MET DST), you wrote in message
<200304201901.h3KJ1fpD029439@pubhealth.ku.dk>:
>Full_Name: Scot McNary
>Version: 1.7
>OS: Windows 98se
>Submission from: (NULL) (134.192.131.246)
>
>
>Hi,
>
>Rgui causes a page fault when asking for lattice graphics after setting
>par(ask=TRUE).  This doesn't happen in 1.6.2, but does in 1.7.0.
>
>library(lattice)
>par(ask=TRUE)
>example(dotplot)
>
>Windows "Details" about crash:
>
>RGUI caused an invalid page fault in
>module R.DLL at 017f:00421ce0.
>Registers:
>EAX=00421cd0 CS=017f EIP=00421ce0 EFLGS=00010206
>EBX=017b9e18 SS=0187 ESP=0096d7b4 EBP=0096d7d8
>ECX=1177ab43 DS=0187 ESI=0096d8c8 FS=8397
>EDX=00000000 ES=0187 EDI=0096d8d0 GS=0000
>Bytes at CS:EIP:
>8b 9a d0 00 00 00 52 d9 7d fa dc 05 c0 1c 42 00 
>Stack dump:
>00000003 00000000 0096d820 01be3200 17b70d3f 0c705027 88080012 01387cf4
017b9e18
>0096d838 11771699 0096d818 0096d820 0096d828 0096d830 01c1e6e8 
>
I don't get this error in Win98 either.  My guess is that it's your
video driver; I don't know why this would happen in 1.7.0 and not
1.6.2, but lattice has changed, and it might be doing something now
that uncovers a bug that was hidden before.  The address of the error
is in R's graphics driver code.

Could you try to see if there's a newer version of your driver
available, or try to see if you can reproduce the bug on a different
machine?  

Another thing that might work is to set options(warn.FPU=TRUE).  This
will enable checking whenever you load a DLL (which many library
packages do) to see if the floating point control word has changed.
If your video driver is messing with it (a common kind of bug), then
you'll likely see warnings popping up.

Duncan Murdoch