R help - Dec 2001 - Corrupt .RData

I'm using R 1.3.1 on a Windows NT 4.0 machine. As you might
guess my machine crashes on me from time to time :-| So far,
forcing a reboot via the Taskmanager has worked reasonably
well: any open R session would enquire politely whether to
save, and do so if requested.

This time, it has not worked out. I get a 'Fatal Error:
unable to restore saved data in .RData' message and
R terminates. Trying to load the file via load() yields 
'Error in load: NewReadItem: unknown type 0' 

Apparently, the system shutdown has hurt the .RData file
somehow. Is there a way of restoring the information at
least partially without too much trouble? As is common
under these circumstances, I have been both working
extensively and not saving sufficiently on this .RData
recently. Any help would be gratefully appreciated.

alexander


--
Alexander Ploner
Applied Statistics
Univ. f. Agricultural Sciences
Gregor Mendel Str. 33
1180 Vienna, Austria, Europe
Tel.: +43-1-47654-5063
E-Mail: ploner at stat.boku.ac.at

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

None, AFAIK.  Seems like you could also use the "auto-save" feature
that I
asked about a while ago.  According to Duncan Temple Lang (Thanks a lot,
Duncan!), he's adding some new features in the development version of R that
will let one do something like running an R command automatically after
every fixed number of top level commands (so you can do a save.image(), say
every 20 commands, for example.

One other preventitive meaure (other than back up often) could be, e.g.,
saving "valuable" objects (large data, or objects that takes a long
time to
reproduce) that will only be referenced rather than modified in separate
files that can be load() or attach().

HTH,
Andy
> -----Original Message-----
> From: Alex Ploner [mailto:ploner at stat.boku.ac.at]
> Sent: Monday, December 10, 2001 9:15 AM
> To: r-help at stat.math.ethz.ch
> Subject: [R] Corrupt .RData
> 
> 
> I'm using R 1.3.1 on a Windows NT 4.0 machine. As you might
> guess my machine crashes on me from time to time :-| So far,
> forcing a reboot via the Taskmanager has worked reasonably
> well: any open R session would enquire politely whether to
> save, and do so if requested.
> 
> This time, it has not worked out. I get a 'Fatal Error:
> unable to restore saved data in .RData' message and
> R terminates. Trying to load the file via load() yields 
> 'Error in load: NewReadItem: unknown type 0' 
> 
> Apparently, the system shutdown has hurt the .RData file
> somehow. Is there a way of restoring the information at
> least partially without too much trouble? As is common
> under these circumstances, I have been both working
> extensively and not saving sufficiently on this .RData
> recently. Any help would be gratefully appreciated.
> 
> alexander
> 
> 
> --
> Alexander Ploner
> Applied Statistics
> Univ. f. Agricultural Sciences
> Gregor Mendel Str. 33
> 1180 Vienna, Austria, Europe
> Tel.: +43-1-47654-5063
> E-Mail: ploner at stat.boku.ac.at
> 
> -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.
> -.-.-.-.-.-.-.-.-
> r-help mailing list -- Read 
> http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
> Send "info", "help", or "[un]subscribe"
> (in the "body", not the subject !)  To: 
> r-help-request at stat.math.ethz.ch
> _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._.
> _._._._._._._._._
> 
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

In response to Alex Ploner's <ploner at stat.boku.ac.at> question
about lost data
from a corrupted .RData, Andy Liaw <andy_liaw at merck.com>
wrote:
> One other preventitive meaure (other than back up often) could be, e.g.,
> saving "valuable" objects (large data, or objects that takes a
long time to
> reproduce) that will only be referenced rather than modified in separate
> files that can be load() or attach().
I can't resist plugging my CRAN package "g.data", which does this
sort of thing
for you, especially if you have many large objects to store but only expect to
use a few at a time.  Essentially:
   R> g.data.attach("/my/storage/area")
to access existing objects (in position 2), and then:
   R> assign("a.valuable.object", big.object, 2)
   R> g.data.save()
to put new ones there.  Better than hundreds of save()'s and load()'s,
IMHO.
-- 
                              -- David Brahm (brahm at alum.mit.edu)
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._