Well, I know for a fact that there are memory leaks in GGobi. They don't
seem serious enough in our experience to warrant a fix though.
Thanks for reporting this. We're trying to be more memory conscious in the
GGobi rewrite.
Michael
On 2/23/07, Ross Boylan <[EMAIL PROTECTED]> wrote:
On Fri, 2007-02-23 at 16:59 -0600, Michael Lawrence wrote:
> How reproducible is this bug? Did you try to rm() the ggobi references
> from the R session?
start R inside ESS
g<-ggobi(data)
4.7M = total reported by interactive xrestop
close(g)
4.6M
rm(g)
4.6M
g2<-ggobi(data)
9.3M
play around with it
71.8 M (! the big jump was when I selected Interaction | brush, though
perhaps it was delayed from earlier...)
close(g2)
71.7 M
rm(g2)
71.7 M
g<-ggobi(data)
76.4 M
interation | brush
76.4 M
display | parallel coords
99.8 M # I did this before the previous brush too, so maybe this
# is the cause. It makes more sense.
close(g)
99.8 M
rm(g2)
99.8 M
gc() # force garbage collect
99.8 M
detach()
99.8 M
library.dynam.unload("rggobi", "/usr/lib/R/site-library/rggobi") #
after many false starts
# I think made no difference, but the R session was busy repeatedly
segaulting.
# When I killed the r process the X resources were freed.
