Re: GetEvent

Rene Brun (Rene.Brun@cern.ch)
Thu, 05 Nov 1998 08:43:40 +0100


Hi Piergiorgio,
This looks like a mismatch between your include files and your code.
You do not identify which version and machine you are using.
I would suggest the usual recipee in this case:
- make clean, etc, build with Root 2.00/13
- check that you can use the browser on your file
- Instrument your code with gObjectTable->Print() to look for
leaks - Check your destructors.

If this does not work, prepare a tar file with the strict minimum code
and a makefile such that we can reproduce the problem.
Send the tar file to rootdev. We would like to restrict roottalk
to communications of general interest.

Rene Brun

Piergiorgio Cerello wrote:
>
> Dear Rooters,
>
> I'm experiencing a strange behaviour of the GetEvent function when
> reading a root file from an interactive session.
>
> The errors are the following:
>
> root [0] .x lmd_example.C
> Start lmd.C Macro
> Path to the main data directory ?
> /data
> Run type ?
> COSM
> Run number (5 digits)?
> 00049
> /data/HDT/COSM00049.hdt file
> 15450 events in HDT file
> Type initial event number
> 600
> Type final event number
> 15449
> Event 600, 184 lmd hits
>
> .....
>
> Event 772, 184 lmd hits
>
> *** Break *** segmentation violation
>
> or
>
> root [0] .x lmd_example.C
> Start lmd.C Macro
> Path to the main data directory ?
> /data
> Run type ?
> COSM
> Run number (5 digits)?
> 00049
> /data/HDT/COSM00049.hdt file
> 15450 events in HDT file
> Type initial event number
> 1
> Type final event number
> 15449
> Event 1, 184 lmd hits
>
> .....
>
> Event 162, 184 lmd hits
> Event 163, 184 lmd hits
> mmuzip: too small source
> Error in <TBasket::Read>: fObjlen = 27932, nout = 0
>
> *** Break *** segmentation violation
>
>
> The producer is writing to file every 250 events, but, if I change this
> value, from the interactive session I get the same errors at different
> event numbers. As I understand, this should mean that the problem does
> not concern the event structure, but something else...
>
> any suggestion?
>
> Thanks in advance, cheers,
>
> Piergiorgio