RE: Mutliple event components

["Torre Wenaus" <wenaus@bnl.gov>]

Hi,
The default (main) component is the header. In going to several components we
are splitting the body into pieces. I've called these pieces clusters. Each
cluster is a cluster of event components that are stored together. They don't
have fixed names because they don't have fixed make-up, but they do need to have
names, assigned when they are created. The name will live in the cluster header
(which references each of the event components in the cluster and is the OID you
use in the retrieval of the cluster). We have to put the cluster names into the
event header too, come to think of it, in order to look up the OIDs.

The user specifies he wants clusters "A" and "B", the OIDs (or filenames, for
non-Objy components!) corresponding to these clusters are looked up in the event
header, and then the OIDs/filenames can be retrieved by GC. Translating from
cluster name to OID/filename deals with the experiment-specific event
architecture and should be on the experiment side of the interface, shouldn't
it?

  - Torre

-- Torre Wenaus, BNL      wenaus@bnl.gov      516-344-4755  Fax 344-4206 --
-- STAR Experiment, RHIC@BNL Computing and software project leader       --
-- B. 510A Room 1-175        http://www.rhic.bnl.gov/STAR/computing.html --
--                           http://www.wenaus.com/torre.html            --



> -----Original Message-----
> From: Henrik Nordberg [mailto:HNordberg@lbl.gov]
> Sent: Wednesday, October 28, 1998 7:20 PM
> To: gcdev
> Subject: Mutliple event components
>
>
> Dear Grand Challenged Ones,
>
> currently we have two event components: the header and the "body" (is
> there a better name for it?). We are now introducing several other
> components. I need to know the names of those and I need to get the
> OID's for them.
> Is it the event header or the "body" that should be treated as the
> default (main) component?
>
> I plan to add an interface to the QE that enables setting the components
> for a query:
>
> void setComponents(string comp);
>
> the string could be a semi-colon delimited sequence of event component
> names. Making the argument a string has the benefit that one can add
> components without updating the QE (and the rest of the SM). The way I
> plan to implement it is to read the component OID's from files named:
> <component name>.oids.bin.
>
> The name of the function could changed (unless you like it). It sets the
> "SELECT" part of our pseudo-SQL, so we could call it: selectComponents()
> or setSelectClause(). Let me know what you prefer.
>
> Regards, Henrik
> _________________________________________
> Henrik Nordberg       <hnordberg@lbl.gov>
> Scientific Data Management Research Group
> Lawrence Berkeley National Laboratory
>