From gwatts@phys.washington.edu Fri Apr 30 14:53:16 2004
Date: Sat, 17 Apr 2004 12:32:44 -0700
From: Gordon Watts <gwatts@phys.washington.edu>
To: bcasey@fnal.gov, d0dfwg@fnal.gov
Cc: d0-conveners@fnal.gov
Subject: RE: common format

Hi Brendan,
  
  I entirely agree. People will be most quickly attracted to the tools
that allow them to do whatever it is they want to do least painlessly.
If the whole B group is using BANA, for example, and I were new and
joining the B group, I'd want to use BANA. All the tools for B physics
would be setup!

  That said, however, the dfwg can't possibly write every tool for
everyone in the collaboration. The goal for the dfwg is to decide and
implement a common format. Once that is done the plan is that the
experiment would centrally produce the root-tuples. This would make it
easier for someone to get started analyzing the data. The hope is that
the more people used it, the more would write tools that would be
generally useful -- the vertexing that you are talking about, for
example.

  We'll probably port some tools over, but we wouldn't, for example,
convert BANA tools to the common format. For example, there is a good
chance we would make sure the bID tools work -- but that will almost
100% likely be me that does the work, both as a member of the bID group
and the dfwg.

  d0root, btw, was created during the switch over from the old
reco-analyze root-tuple. If you look at its objects closely you'll see
that it is similar to tmb tree. If we end up with a format similar to
tmb tree one thing on the table is to try to combine the two.

  Finally, some of the questions we sent around in the questionnaire
were aimed at finding out why people use what they use. So far I've seen
two answers:

- The tools are available.
- Historical

      Cheers,
          Gordon.


-----Original Message-----
From: bcasey@fnal.gov [mailto:bcasey@fnal.gov] 
Sent: Friday, April 16, 2004 10:56 PM
To: d0dfwg@fnal.gov
Cc: d0-conveners@fnal.gov
Subject: common format

Herb et al,

I raised a point at the conveners meeting a few weeks ago about the
charge of your group and it seems to have been forgotten so I'd like to
raise it again.

I think for your working group to be successful, it must figure out why
people are using different formats and solve that with one format.

I think the main reason different formats exist is so that people can
run high level algorithms with their own choice of inputs that only run
in a particular format or only accept inputs in a particular format.
Specifically kinematic fitters.

For instance, in a lifetime analysis, it makes much more sence to first
reconstruct a particle, find its secondary vertex, then find a primary
vertex explicity excluding secondary tracks from the primary vertex,
than to try and use vertex information computed globally for the entire
event.  This requires a vertexing package.  There is no supported
vertexing package included in the tmb_tree format.  So people who do
root based lifetime studies move to the d0root format.

So if these tools or algorithms are not incorporated in the new format,
people will simply convert the new format to the more familiar formats
so that they can use the existing tools.

I think the success of this endevor completely relies on incorporating
all these user level tools into the single format.

By incorporating, I mean a user would add one #include statement to
their analysis program to access a tool, not have to check out half the
Chicago phone book worth of code and compile it in their own area to
access it.

Brendan