Re: root file basket size

From: Victor Perevoztchikov (perev@bnl.gov)
Date: Wed Jan 05 2000 - 17:59:22 MET


Hi Rene,

> The first thing to check is to read your Trees(s) using the standard
> Root browser. 

Just in browser we have a crash:

In reading in TBasket::Streamer we have

 b >> fBufferSize    // fBufferSize == 64K
 b >> fLast          // fLast == 68K
 new TBuffer with fBufferSize == 64K
 b.ReadFastArray(new buffer, fLast)
 so it copyes 68K array into 64K buffer

So it looks like that during writing buffer correctly expands
but its size is not written correctly. Thats why in reading
we have fLast > fBufferSize 




Rene Brun wrote:
> 
> Hello Bill,
> The basket size is automatically extended to accomodate at least
> one entry. I don't think there is a problem there.
> 
> The first thing to check is to read your Trees(s) using the standard
> Root browser. If you have only TClonesArrays, you should be able
> to inspect all your attributes with the standard Root without your
> classes. If this works, you should then suspect a possible error
> in the way you set your branch addresses in the reading program.
> 
> Let me know when you have a (small) file with problems, I will
> investigate.
> 
> Rene Brun
> 
> William Love wrote:
> >
> >    Root gurus
> >
> >    I am using Root 2.23/09 and writing a TTree with three
> >  TClonesArrays, Tracks, Hits, and Pixels.  Recently I
> >  suffer unreproducable troubles with root output files.
> >  Though the  tree is well behaved while I am running
> >  the writing program, if I close the file and exit and then
> >  reopen the file I am often stuck with a segmentation
> >  fault in trying to look at the TTree, sometimes accompanied
> >  by a message about insufficient memory.
> >    Victor (Perevoztchikov) says he remembers a rootalk
> >  dialog about buffer (basket) size problems.  My events
> >  have usually about 16000 Hits, which means members
> >  would come near the 64 Kbytes basket size.  Does root
> >  maybe have a problem when variables just overflow a
> >  basket?
> >    I would love to send an example script but I have had
> >  runs of 200 events that work fine and runs of 5 events
> >  that produced an unusable root file.  Its frustrating.
> >  Any information or advice would be appreciated.
> >                                    Thanks
> >                                     Bill Love

-- 
Victor M. Perevoztchikov   perev@bnl.gov  perev@vxcern.cern.ch       
Brookhaven National Laboratory MS 510A PO Box 5000 Upton NY 11973-5000
tel office : 631-344-7894; fax 631-344-4206; home 631-345-2690



This archive was generated by hypermail 2b29 : Tue Jan 02 2001 - 11:50:16 MET