R devel - Oct 2002 - possible memory leak in 1.6 (PR#2160)

Full_Name: Anne-Laure Boulesteix
Version: 1.6.0
OS: Linux
Submission from: (NULL) (141.84.161.156)


Hi,

I noticed the following problem with 1.6.0 (which was not present in 1.5.1): 

I work with very large datasets and encountered a problem with memory
allocation,
(probably a memory leak).  I wrote a programm with a loop.  In this loop in
each
iteration I create a  very large matrix and assign it to a variable which
is the SAME for all the iterations.  So at each round the new matrix overwrites
the old  matrix.  In 1.5.1  the garbage collector seems to clean up the old
unused memory.  Unfortunately, in 1.6.0 with each round the memory used by R
increases until the R process is killed. 

Is there any way around this?

Many thanks and best regards,

  A.L. Boulesteix

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel 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-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

socher@stat.uni-muenchen.de writes:
> Full_Name: Anne-Laure Boulesteix
> Version: 1.6.0
> OS: Linux
> Submission from: (NULL) (141.84.161.156)
> 
> 
> Hi,
> 
> I noticed the following problem with 1.6.0 (which was not present in
1.5.1):
> 
> I work with very large datasets and encountered a problem with memory
> allocation,
> (probably a memory leak).  I wrote a programm with a loop.  In this loop in
> each
> iteration I create a  very large matrix and assign it to a variable which
> is the SAME for all the iterations.  So at each round the new matrix
overwrites
> the old  matrix.  In 1.5.1  the garbage collector seems to clean up the old
> unused memory.  Unfortunately, in 1.6.0 with each round the memory used by
R
> increases until the R process is killed. 
> 
> Is there any way around this?
Example code that would allow someone to reproduce and hunt down the
problem would help. This could be the same issue that Achim Zeileis
reported a few days ago. In that case the current source snapshots may
have it fixed.

Please be specific when you report problems. A broad outline of your
program is not sufficient to let us guess where the problem lies.
We could write a minimal loop which fits your description, but so
could you, and you're much closer to telling whether it is actually
displaying the same problem.

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard@biostat.ku.dk)             FAX: (+45) 35327907
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel 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-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

On Mon, 14 Oct 2002 socher@stat.uni-muenchen.de wrote:
> Hi,
>
> I noticed the following problem with 1.6.0 (which was not present in
1.5.1):
>
> I work with very large datasets and encountered a problem with memory
> allocation,
> (probably a memory leak).  I wrote a programm with a loop.  In this loop in
> each
> iteration I create a  very large matrix and assign it to a variable which
> is the SAME for all the iterations.  So at each round the new matrix
overwrites
> the old  matrix.  In 1.5.1  the garbage collector seems to clean up the old
> unused memory.  Unfortunately, in 1.6.0 with each round the memory used by
R
> increases until the R process is killed.
Yes, it's a bug (though it isn't the matrix that leaks).  It's
probably been fixed.

You could download the current `patched' version and try that.

	-thomas

Thomas Lumley			Asst. Professor, Biostatistics
tlumley@u.washington.edu	University of Washington, Seattle

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel 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-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._