Re: [Brahms-dev-l] NOTE about using PROOF (bug?)

From: flemming videbaek <videbaek@rcf.rhic.bnl.gov>
Date: Wed Jun 08 2005 - 09:59:08 EDT
As far I know,
pROOF is a parellel system, and as I understand the only objects that are
automatically summed
at the end of a session are histograms (either in your MakeDst or in the
Draw commands).
Thusin you you will have to use a hsitogram.

/fv
----------------------------------------------------------------
Flemming Videbaek
Physics Department
Brookhaven National Laboratory

e-mail: videbaek@bnl.gov
phone: 631-344-4106
----- Original Message ----- 
From: "Johnson, Erik B" <ebj@ku.edu>
To: "Hironori Ito" <hito@rcf.rhic.bnl.gov>; "devlist"
<brahms-dev-l@lists.bnl.gov>
Sent: Wednesday, June 08, 2005 9:31 AM
Subject: RE: [Brahms-dev-l] NOTE about using PROOF (bug?)


>
> I have a question about proof.
> I want to sum up a variable over all events using proof, but for some
reason I get a zero at the end.
>
> I generated a class using the TTree::MakeSelector function.
>
> In the Begin function I set a variable to zero, which I plan to store the
sum.
> In Process I add 1 to the each valid event.
> In Terminate, I print out the value of the variable and it is zero.
>
> I know this variable is increasing in value, but for some reason I end up
with zero.  Do I need to think of a proof process as actually 14 different
processes, where in the end I have to sum the results from each process?  I
didn't think so, but I'm not entirely sure how this system works.  So any
help would very much appreciated.
> Thanks,
> Erik
>
>
>
> -----Original Message-----
> From: brahms-dev-l-bounces+ebj=ku.edu@lists.bnl.gov on behalf of Hironori
Ito
> Sent: Tue 6/7/2005 2:07 PM
> To: devlist
> Cc:
> Subject: [Brahms-dev-l] NOTE about using PROOF (bug?)
> Hello.  Since some of you use PROOF in rcas, I thought I write a few
> warning (or some strange thing (bug???)) that I noticed while I use it.
>
> 1.  PROOF goes nowhere (sometimes) when you try to draw  non-existing
> variable.  It (sometimes) did not return the prompt and just sit there.
> I had to kill the process.
> 2.  PROOF does not like to merge empty histograms.
> 3.  When you kill your local root/proofserv, it does not necessary kill
> proofserv process at other machines.  To kill, you might need to kill
> proofserv at the master machine.
>
> Hiro
>
>
> _______________________________________________
> Brahms-dev-l mailing list
> Brahms-dev-l@lists.bnl.gov
> http://lists.bnl.gov/mailman/listinfo/brahms-dev-l
>
>
>
>
> _______________________________________________
> Brahms-dev-l mailing list
> Brahms-dev-l@lists.bnl.gov
> http://lists.bnl.gov/mailman/listinfo/brahms-dev-l
>


_______________________________________________
Brahms-dev-l mailing list
Brahms-dev-l@lists.bnl.gov
http://lists.bnl.gov/mailman/listinfo/brahms-dev-l
Received on Wed Jun 8 10:00:21 2005

This archive was generated by hypermail 2.1.8 : Wed Jun 08 2005 - 10:00:28 EDT