[R] Hangup on save.image () / q ()

Damon Wischik djw1005 at cam.ac.uk
Tue Sep 2 18:38:12 CEST 2003

> I am doing some data handling and tabualtions and then I am using 
> glmmPQL to fit a binomial hierarchical model (which I assign to a
> new object). If I do a save.image () or a q ('yes') after this, I
> get the following warning messages: 
>   Warning messages: 
>   1: namespaces may not be available when loading 
>   2: names in persistent strings are currently ignored 
> Sometimes R seems to shut down properly, sometimes R hangs up during 
> the quit, in any case R cannot be restarted after this because the
> .RData seems to be corrupt.

I have trouble with closing R under ESS.  I have this trouble _every_time_
I plot any graphics at all. I've had the same trouble under several
Windows 2000 and Windows XP machines, under both Emacs and Xemacs. I don't
know why other people don't have the same trouble. I don't have any
trouble when I don't plot any graphics. (But then, I never do
save.image()). Do you plot any graphics in your R session? 

I posted messages to the r-help and ess-help mailing lists:

>From r-help, I was told that it's a Windows/Emacs issue, and not a topic
for r-help. From ess-help, I was told that it's a Windows/Emacs issue, one
which "either R can fix or Emacs can fix", and not a topic for ess-help.
It's something to do with relinquishing file handles, and "it probably
won't get fixed anytime soon".

I assume we're in a minority. Otherwise, the R FAQ for all platforms
should give a warning rather than just saying
  6.2 Should I run R from within Emacs?
  Yes, _definitely_.
and the R FAQ for Windows would mention it.

The only question left is: how do you cope with it? I still use ESS, and I
close my sessions by switching off Emacs, which then has Windows do the
dirty business of shutting down R, because this way I don't have to put up
with unresponsive windows. Of course, this is no help to you, as you want
to save your data...

Damon Wischik.

More information about the R-help mailing list