Present: Herbert Greenlee, Gianluca Petrillo, Brian Rebel
Connected: Tingjun Yang (for the second topic only)
Extension of geometry to accommodate shared readout channels: wrap up
The proposal of storing an invalid
WireID in a hit, unless the wire is unambiguously determined, has been accepted.
Work will be needed to change the current behaviour of most (or all) the hit finding algorithms.
Channel filtering
The proposal in the material was described.
It was agreed to proceed according to its content, and in particular:
-
having a two-layer model separating framework interface and service provider
-
having an abstract service interface, and many experiment-specific implementations
-
having a default implementation provided by LArSoft with a simple input format (FHiCL)
-
in general, following the model of the DB access from e.g. pedestal retrieval and the coming
DetectorProperties adaption to DB
In addition to that:
-
[Brian Rebel] provide the ability to read information from a Comma-Separated Values (CSV) text source, possibly as a special case of the database access
-
after discussion, no urgent need has been expressed for reading the same information from the event itself, although it might turn out to be useful in cases where remote nodes are unable to connect to the DB
-
iterators have been commented to be potentially useful, unessential and possibly "overkill"
-
bit map interface has no current use case and should therefore not be pursued at this time
In conclusion:
-
LArSoft will pursue the model proposed in the slides
-
the implementation of DB-accessing services is left to the experiments who need it
-
we will use caution not to prevent a future extension reading information from the event unless necessary
-
the old
ChannelFilter will be temporary supported, deprecated, and finally removed
-
none of the optional features mentioned in slide #9 will be pursued at this time
-
the time scale of this implementation is one month
There are minutes attached to this event.
Show them.