Hi Tome,
OK, I understand the problem now. It is fixed in the CVS source.
Thanks for reporting it.
Rene Brun
On Sat, 3 Nov 2001, tome anticic wrote:
>
> Hi Rene,
>
> The memory leak is about 800 bytes/iteration. Not much,
> but my monitoring program has to continuously generate plots, full time,
> so this is unfortunately a problem. I have observed the same effect on
> separate machines (my RH6.2 laptop with 3.02./02, and on a 6CPU HP
> netserver running RH7.1 with 3.01/06), so it is not some local error (but
> one never knows...).
>
> However, if printing gifs instead of eps, this problem dissapears!
> Unfortunately I am running in batch mode, so I cannot use this.
>
> Tome
>
>
> > I cannot reproduce this problem.
> > Could you tell me what is the estimated memory leak per iteration ?
> >
> > About your second question. If you run interactively,
> > you can produce a gif file directly by replacing
> > c1->Print("xxx.eps");
> > by
> > c1->Print("xxx.gif");
> >
> > Rene Brun
> >
> > On Sat, 3 Nov 2001, Tome Anticic wrote:
> >
> > >
> > > Thanks for the quick reply.
> > >
> > > Unfortunately, I tried it under root 3.02/02 and the problem persists.
> > > gObjectTable->Print() reports no problems however. But the
> > > memory usage (using top) keeps increasing on each iteration
> > > of the Print() method...
> > >
> > > Actually, I also have a related question. The reason I
> > > print the postscript file is to convert it to gif (via
> > > pstopnm and ppmtogif). This is a slowish process, and I
> > > was wondering if there is a faster way to get a gif of
> > > a canvas?
> > >
> > > Tome
> > >
> > >
> > > On Fri, 2 Nov 2001, Rene Brun wrote:
> > >
> > > > Hi Tome,
> > > >
> > > > This problem has already been fixed.
> > > > Move to a more recent version.
> > > >
> > > > Rene Brun
> > > >
> > > > Tome Anticic wrote:
> > > > >
> > > > > Hi,
> > > > >
> > > > > I would really appreciate if somebody could tell
> > > > > me what I am doing wrong.
> > > > >
> > > > > I need to generate a lot of plots at short intervals.
> > > > > So I do a graph in a canvas and use the Print() method
> > > > > to print a postscript file . But I noticed that the
> > > > > Print() method causes an increasing amount of memory to be
> > > > > used. How can I prevent this and free the memory?
> > > > > Below is a sample code where
> > > > > this can be observed (do top from a shell and watch...)
> > > > >
> > > > > Any help would be appreciated. I use root 3.00 /egcs 2.91.66
> > > > >
> > > > > Thanks,
> > > > > Tome
> > > > >
> > > > > void testMemory()
> > > > > {
> > > > > Float_t X[100];
> > > > > Float_t Y[100];
> > > > >
> > > > > TCanvas *c1;
> > > > > c1 = new TCanvas("c1","canvas");
> > > > >
> > > > > for (Int_t i=0; i< 100;i++) {
> > > > > X[i] = i;
> > > > > Y[i] = i;
> > > > > }
> > > > > TGraph* grh = new TGraph(100,X,Y);
> > > > > grh->Draw("AL");
> > > > >
> > > > > while (1) {
> > > > > cout << " do plot " << endl;
> > > > > c1->Update();
> > > > > // this causes memory to be used and NOT freed
> > > > > c1->Print("xxx.eps");
> > > > > }
> > > > > }
> > > >
> > >
> >
>
> --
>
> Tome Anticic
> Rudjer Boskovic Institute
> Department of Experimental Physics
> Bijenicka 54
> 10000 Zagreb
> CROATIA
> Phone: 385-1-4561-028
> Fax: 385-1-4680-239
>
This archive was generated by hypermail 2b29 : Tue Jan 01 2002 - 17:51:06 MET