Re: [ROOT] Pointer to array of TH1F. Filling of under- and over flows ???

From: Rene Brun (Rene.Brun@cern.ch)
Date: Tue May 08 2001 - 10:51:20 MEST


Hi Martin,

You are deleting your histogram after pMult_[i] = *pDum; 
You then have an illegal access when you attempt to Fill it.


Rene Brun


Martin Karlsson wrote:
> 
> Dear Rooters,
> 
> I'm having problems with filling of TH1F.
> 
> I'm working on an application in which a bunch of histograms are declared as
> 'pointer to array'.
> Before, I used v. 2.23/12 and it worked without any problems; now, however I
> have upgraded to version
> 3.00/06 and then when I run the program it crashes (seg. violation). It turns
> out that the crash occurs when a Fill() call is done and it is an over- or
> under flow that is to be filled.
> 
> The following test program illustrates the problem:
> ////////////////////////////////
> int main()
> {
>   TROOT litta("try","try ");
> 
>   TH1F  simp;
>   TH1F * pMult_=NULL;
> 
>   char hname[100], htitle[100];
> 
>   pMult_ = new TH1F[2];
>   for( Int_t i = 0; i < 2; i++)
>     {
>       sprintf( hname, "pMult_%d", i);
>       sprintf( htitle, "pointer to Mult %d", i);
>       TH1F * pDum = new TH1F(hname, htitle, 11, -0.5, 10.5);
>       pMult_[i] = *pDum;
>       delete pDum;
>     }
> 
>   simp = TH1F("simp","simp", 11, -0.5, 10.5);
> 
>   for(Int_t i = 0; i < 13; i++)
>     {
>       pMult_[0].Fill(i);
>       pMult_[1].Fill(i);
>     }
> 
>   delete [] pMult_;
> 
>   return 0;
> }
> /////////////////////////////
> 
> If I compile this (with the --new option sent to root-config (as I'm using a
> TMapFile in my application)) and then run it, the Fill() call with under- or
> overflow will crash (seg violation) the program.
> 
> I can get the small test program to work if I do one of the following:
> *compile WITHOUT (sic) -- new flag
> *just run as a macro
> *use v. 2.23/12
> *remove the non-pointer-TH1F simp
> *change the number of bins to say <11 or > 12 !!!
> 
> So, I'm a bit confused about this behavior... of course I could re-declare all
> my histograms but that's not fun and I  do like to know what I'm doing wrong.
> So if anyone has a clue I would very much appreciate being enlightened.
> 
> Best Regards,
> Martin
> 
> P.S. I use RH 6.1 on a SMP i686 (two processors) and with egcs-1.1.2
> 
> 
> 
> --
> Martin Karlsson
> Department of Nuclear Physics, University of Lund
> Box 118, SE-221 00 Lund, Sweden
> Phone: +46 (0)46 222 96 77  or +46 (0)736 87 84 15
> Fax: +46 (0)46 222 76 29
> 
>



This archive was generated by hypermail 2b29 : Tue Jan 01 2002 - 17:50:44 MET