[rkward-devel] rkward fatal crash

[Dan McCloy]

update: something was wrong with my Cairo package (not sure what) such
that every call to CairoPDF() was crashing.  re-installing Cairo
solved the problem.  Sorry to have troubled you.
-- dan

On Sun, Nov 6, 2011 at 09:25, Dan McCloy <drmccloy at uw.edu> wrote:
> reproduced within RKWard, and similar crash when R run in the terminal (with
> slightly different error message, see below).  I'll contact Simon.  thanks.
>
>  *** caught segfault ***
> address (nil), cause 'memory not mapped'
>
> Traceback:
>  1: strwidth(legend, units = "user", cex = cex)
>  2: legend("topleft", legend = leg, bty = "n", inset = 0.02, cex = 0.75)
>
> Possible actions:
> 1: abort (with core dump, if enabled)
> 2: normal R exit
> 3: exit R without saving workspaceg
> 4: exit R saving workspace
>
> On Sun, Nov 6, 2011 at 04:55, Thomas Friedrichsmeier
> <thomas.friedrichsmeier at ruhr-uni-bochum.de> wrote:
>>
>> Hi,
>>
>> On Sunday 06 November 2011, Dan McCloy wrote:
>> > What I was doing at the time:
>> >
>> > Had just been tweaking a set of seven graphs to plot in a 4x2 layout
>> > using
>> > par(mfcol=).  The graphs were fairly simple scatterplots with Lowess
>> > curves.  I had gotten all the axes and cex values to where I wanted
>> > them,
>> > and closed the X11 graph window and ran the whole set of graph commands
>> > again with a CairoPDF() and dev.off() wrapper.  That's when the crash
>> > occurred.  The PDF was initialized but remains at zero bytes.  The
>> > specific
>> > code is included as an attachment.  Note that I had the following
>> > packages
>> > loaded: Cairo, Hmisc.
>>
>> [...]
>>
>> > let me know if i can be of further help in tracking down the source of
>> > this
>> > error.
>>
>> thanks for your detailed bug report.
>>
>> I don't have the data objects needed to run the example. I inserted some
>> dummy
>> plots, instead, but that did not cuase any problems.
>>
>> This crash might have been RKWard's fault, or it might have been a crash
>> inside Cairo. Since, as far as I understand, no on-screen device was
>> active at
>> the time, I believe it is rather likely to be the latter.
>>
>> Here are some things for you to test:
>> - If you run the same code again, does the crash happen, again?
>> - If you run the same code in a "plain" R session on the terminal, what
>> happens?
>>
>> If you can reproduce the bug, *and* you can also reproduce it in a
>> terminal R
>> session, then I suggest to report the bug to the maintainer of the Cairo
>> package (Simon Urbanek). If you can reproduce the bug, but only inside
>> RKWard,
>> it would be very helpful, if you could send me the data objects.
>>
>> If you cannot reproduce the bug, then unfortunately, there is not much
>> anybody
>> can do about it.
>>
>> Regards
>> Thomas
>
>