Minutes of the off-line meeting,
1st September 1998
Present: P.Binko, T.O'Reilly,
M.Cattaneo, J.Saborido, A.Tsaregorodtsev, T.Cass, F.Ranjard, G.Gracia
Minutes: A.Tsaregorodtsev
Agenda:
News
A quick UML introduction (T.O'Reilly)
First ideas on OO Track finding and fit (G.Gracia)
AOB
1. News
There will be a SW meeting on Wednesday, 2 September, to discuss class
design in the Track domain. More people are invited for the discussion (Pere,
others).
Iain will make a status report on SICB C++ classes on the next meeting.
Rational Rose student license is available on NT for eveluation. It has
a limit of no more than 30 classes per project.
Glossary: Marco distributed a list of terms
for top level data structure elements to be considered and fixed after a discussion in the
group.
Pavel will report on the next meeting on refinement of the OO event data
model following discussions after his first presentation.
2. A quick UML introduction (T.O'Reilly)
Thomas showed basic elements of the UML notation used to draw class
diagrams in Rational Rose and other CASE tools (see slides, ppt). This is a notation which we have adopted for
OO design purposes. The tutorial material will be available on the WWW as a
permanent page.
3. First ideas on OO Track finding and fit (G.Gracia)
Gonzalo presented requirements for the data necessary to perform the track
finding and fit (see slides, ppt). He identified among clients of the
tracking software subdetectors, analysis, event display, trigger, etc. In a first attempt
to define possible track related classes and their relations he identified a basic Track
class with MCTrack and FittedTrack derived from that as subclasses. FittedTrack contains
ErrorMatrix object in addition to a basic Track. It also contains a list of Nodes at which
the track parameters are evaluated during the fitting procedure. A separate TrackManager
class is suggested to contain steering of the track finding/fitting procedure (like
getting seeds for track finding, invoking fitting algorithm, etc). TrackManager object
uses a TrackFitter object which encapsulates the track fitting algorithm.
In a discussion that followed Pavel stressed three points that
should be kept in mind while designing a class diagrams:
Life time of participating objects should be taken into account. Mix
with care objects with event scope and run scope;
Object relations should be constructed with due account to a
sequence of objects ctreating (who creates what);
Objects likely to be stored persistently should be identified and shoudl
not depend on purely transient objects.
4. A.O.B.
Pavel raised the question of which top level event structure is more
appropriate: One which contains DataType (like MC, Raw, Reconstruction) level under the
Event box or a Subdetector second level. It was a general feeling that the first one is
more acceptable. This is also a choice in BaBar, CMS and some other experiments. The
decision was taken to proceed with the first model. Pavel will present more elaborated
version of it on the next meeting.
The next meeting will be on 8th September at 10:30 in
32-SC-22 as usual, with the following agenda
LHCb Home | Computing Home | Search
|