FWD Software

[Akio Ogawa <akio AT bnl.gov>]

Hello

Well... It looks like we do have correct and enough info, if we take 
GetLast() and dont look at elements of the array beyond GetLast(), 
ignoring what GetEntries() or GetSize() tells you.

But file size may be largfer than needed to be.

OGAWA, akio                 Email : akio AT bnl.gov
BrookHaven National Lab. Physics Dep. 510A 2-235
Upton, NY, 11973-5000          Tel. 631-344-5293

On Mon, 20 Nov 2023, Brandenburg, Daniel wrote:

> Hi Akio,
>
>  
>
> Thanks, but I am not sure I follow. What do you need from me?
>
>  
>
> Also, does this mean MuDst are corrupt beyond repair? Can this info be 
> correctly rebuilt from more fundamental info
> saved in MuDst?
>
> Otherwise we may need to fix this before the imminent production.
>
>  
>
> Best,
>
> Daniel Brandenburg
>
> Assistant Professor
>
> College of Arts and Sciences 
>
> Department of Physics
>
> Ohio State University
>
>  
>
>  
>
>  
>
> From: Akio Ogawa <akio AT bnl.gov>
> Date: Monday, November 20, 2023 at 11:36 AM
> To: Weibin Zhang <weibinz AT ucr.edu>
> Cc: Brandenburg, Daniel <brandenburg.89 AT osu.edu>, 
> star-fwd-software-l AT lists.bnl.gov
> <star-fwd-software-l AT lists.bnl.gov>
> Subject: Re: Problem with StMuFcsCluster
>
>
>
> Hello
>
> Please see modified /star/u/akio/forWeibin/test.C
>
> It seems
>
>     int nl = hits->GetLast();
>
> gives "correct" number of hits minus 1. It looks like GetSize() and
> GetEntries() give current size of TRefArray which incldes allocated
> but not currently used elements.
>
> Note that loop need to be
>
>    int nl = hits->GetLast();
>    for (int i=0; i<=nl; i++){}  //include i==nl!!!
>
> to loop over all hits.
>
> There is TRefArray::Compress(). But it makes GetLast() point to same
> place -1 as GetSize/GetEntries, not otherway around.
>
> Removing each by hand
>    if(nl+1 < nh){ for(i=nl+1; i<nh; i++) {hits->RemoveAt(i);} }
> makes GetEntries() consistent with GetLast() but GetSize() remains same.
>
> I'm not sure how TRefArray are added/reset/reused/cleared in Mudst and
> if this is what Daniel expects to see... Or if this is optimize for file
> size?
>
> OGAWA, akio                 Email : akio AT bnl.gov
> BrookHaven National Lab. Physics Dep. 510A 2-235
> Upton, NY, 11973-5000          Tel. 631-344-5293
>
> On Fri, 17 Nov 2023, Weibin Zhang wrote:
>
> > Hi, Akio and Daniel:
> > I found that the MuDst FCS cluster is not properly filled with 
> corresponding hits, specifically, one cluster has
> > hits that have different detector Ids.  Here is a simple macro to reproduce 
> this problem:
> >
> > macro:  /gpfs/mnt/gpfs01/star/pwg/weibin/DY/test.C
> > and the MuDst file: /star/data01/pwg/weibin/DY/condor/pi-/pi_0.MuDst.root
> >
> > Please check it.
> >
> > Best,
> > Weibin
> >
> >