[R-SIG-Mac] problems with the data editor

[John Fox]

Dear Simon,

> -----Original Message-----
> From: Simon Urbanek [mailto:simon.urbanek at r-project.org]
> Sent: January-09-09 3:01 PM
> To: John Fox
> Cc: R List
> Subject: Re: [R-SIG-Mac] problems with the data editor
> 
> John,
> 
> thanks for the report.
> 
> On Jan 9, 2009, at 14:07 , John Fox wrote:
> 
> > Dear list members,
> >
> > I thought that I recalled a recent posting reporting a problem with
> > the R
> > data editor, but I haven't been able to locate the message in the list
> > archive.
> >
> > In any event, I can reliably cause the R.app data editor either to
> > hang or
> > to produce errors in a couple of ways, the first of which is easier to
> > describe (the X11 data editor seems more robust, if limited in its
> > capabilities):
> >
> > (1) Edit an empty data frame (which works in the Windows and X11 data
> > editors): edit(as.data.frame(NULL)).
> >
> > (2) Edit any data frame "vigorously," adding a deleting rows and
> > columns
> > repeatedly. E.g., library(car); edit(Prestige).
> >
> > I'm using R 2.8.1 installed from the binary on CRAN under OS X
> > 10.5.5 on a
> > MacBook.
> >
> 
> Can you, please, send me the crash reports you have? I can reproduce
> some of it, but I wonder if it's caused by the same problems as your
> crashes.

The crash produced by editing an empty data frame is simply for R to hang --
spinning ball, with an error message: SET_VECTOR_ELT() can only be applied
to a 'list', not a 'NULL'. I have to force-quit from R.

Simply editing a data frame "vigorously" has produced different errors,
including hangs, but what happens seems to be variable. I just was able to
generate the following:

*** caught bus error ***
address 0x6eb, cause 'non-existent physical address'

Traceback:
 1: edit.data.frame(Prestige)
 2: edit(Prestige)

Possible actions:
1: abort (with core dump, if enabled)
2: normal R exit
3: exit R without saving workspace
4: exit R saving workspace
Selection:

> 
> 
> > I wonder whether anyone else has observed these problems or has an
> > idea
> > about how to fix them.
> >
> 
> The data editor is horribly old code (or rather horrible, old code)
> that desperately needs re-writing. The plan was originally to do that
> for GUI 2.0, but it is not in sight due to the lack of manpower. I can
> at least see if I can fortify it a bit better so it doesn't generate a
> crash (I was hoping I won't have to do that since it's really an ugly
> mess). FYI the nightly builds won't run the next three nights, because
> we are having a complete shutdown here for maintenance until Monday,
> so the R.app binaries won't be fixed until Tuesday.

I'm sorry to make work for you. I think that it's probably better not to
have a data editor than to have one that fails frequently. I don't use the
data editor much myself, but new users in particular are put off by the
instability, I believe.

> 
> Thanks,
> Simon
> 
> 
> > (I'm copying Brian Ripley since we exchanged emails about this issue
> > earlier
> > today.)
> >
> 
> (Was it a BCC? I didn't see it in the e-mail ...).

This is curious, since I've just confirmed that the message in my out-going
mailbox had Brian as a CC. Oh well.

Best,
 John

> 
> 
> > Regards,
> > John
> >
> > ------------------------------
> > John Fox, Professor
> > Department of Sociology
> > McMaster University
> > Hamilton, Ontario, Canada
> > web: socserv.mcmaster.ca/jfox
> >
> > _______________________________________________
> > R-SIG-Mac mailing list
> > R-SIG-Mac at stat.math.ethz.ch
> > https://stat.ethz.ch/mailman/listinfo/r-sig-mac
> >
> >