Hi Philippe,
Thank you, I think I will move to ACLiC finally, because this little
flaw and some other quirks have broken my program.
Stanislav.
On Fri, 2 Aug 2002, Philippe Canal wrote:
> Hi,
>
> CINT has a known defect in its optimization code that can cause the type of
> effect you notice when you create object on the stack within a loop (and
> maybe a few more nesting).
>
> To work-around the problem, move the declaration: TLorentzVector tmp; //
> Local declaration
> outside of the loop.
>
> Cheers,
> Philippe.
>
> -----Original Message-----
> From: owner-roottalk@pcroot.cern.ch
> [mailto:owner-roottalk@pcroot.cern.ch]On Behalf Of Stanislav NESTEROV
> Sent: Friday, August 02, 2002 11:01 AM
> To: Pasha Murat (630)840-8237@169G
> Cc: roottalk@pcroot.cern.ch
> Subject: Re: [ROOT] SIGSEGV in TLorentzVector addition !!
>
>
>
> Thank you Pasha,
>
> I think it may be the best way, BUT if I do so, everything
> is working FINE. I want to understand why in CINT session script isn't
> interpreted well.
>
> Best regards,
> Stanislav
> On Fri, 2 Aug 2002, Pasha Murat (630)840-8237@169G wrote:
>
> > Stanislav: try to compile your code using ROOT script compiler and
> > debug it with the symbolic debugger. This procedure normally
> > takes less than 5 minutes and gives direct answer about what the
> > problem is.
> > -best, Pasha
> >
> > Stanislav NESTEROV wrote:
> > >
> > > Hi Rene,
> > > I've try to run script that uses TLorentzVector and TVector3, and in
> > > some very simple cases this script is aborted via segfault.
> > >
> > > This is it (at the end it crashes on tmp=vpk[0]+vpk[3])
> > >
> > > To run this script see
> > > "/afs/cern.ch/user/n/nesterov/public/"
> > > for neccessary files (pi_nodedx_all.root).
> > > ----------------------------------------------------------------------
> > > #define PIM 139.56995*1e-3
> > > #define KM 493.677*1e-3
> > > #define PZM 134.9764*1e-3
> > > #define NTPNUM "h96"
> > > void testLorentz(){
> > > gSystem->Load("libPhysics.so");
> > > Double_t M[4] = {KM,KM,KM,KM};
> > > Double_t M1122[4] = {KM,KM,PIM,PIM};
> > > TFile f("pi_nodedx_all.root");
> > > TTree * tree = (TTree*)f.Get(NTPNUM);
> > >
> > > if (!tree) {cout <<"Bad luck"<<endl; return;}
> > > Float_t ptpi[4],thtpi[4],phpi[4];
> > > tree->SetBranchAddress("Ptpi",ptpi);
> > > tree->SetBranchAddress("Phipi",phpi);
> > > tree->SetBranchAddress("Thtpi",thtpi);
> > > TVector3 p[4];
> > > TLorentzVector v[4],g,vpk[4];
> > > for (Int_t i=0;i<tree->GetEntries();i++)
> > > {
> > > if (i%1000 == 0 ) cout <<i<<endl;
> > > tree->GetEntry(i);
> > > for (Int_t j=0;j<4;j++) // Setting vectors;
> > > {
> > > p[j].SetPtThetaPhi(TMath::Abs(ptpi[j]),thtpi[j],phpi[j]);
> > > v[j].SetVectM(p[j],M[j]);
> > > }
> > > g = v[0]+v[1]+v[2]+v[3];
> > >
> > > Float_t lb=g.M();
> > > if ((lb<2.11)) {
> > > TLorentzVector tmp; // Local declaration
> > > for (Int_t j=0;j<4;j++) {
> > > vpk[j].SetVectM(p[j],M1122[j]);
> > > }
> > > tmp = vpk[0]+vpk[3]; // Seg. fault!!!!!!!!!
> > > cout<<"Success"<<endl;
> > > }
> > >
> > > }
> > > }
> > > ---------------------------------------------------------
> > >
> > > 1) If I moved `tmp' declaration outside if(lb<2.11) scope, everything is
> > > OK.
> > >
> > > 2) If I go into debug mode and jump onto local declaration -- everything
> > > is OK
> > >
> > > 3) If I don't read tree values --- everything is OK.
> > >
> > > It's just mistics!!
> > >
> > > Best regards,
> > > Stanislav.
> >
>
>
>
This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:51:03 MET