Next: Processors: Definitions and Rules
Up: Processors: A Discussion
Previous: Maintaining Data Structure Integrity
Contents
Sadly, the answer is ``No''. Not all the SUs that are called processors follow
the model of a process that modifies a data structure in some reversible way.
For example:-
- Standard input and output are considered to be processors but they don't
modify the data structure, they simply move it between memory and an external
medium.
- A filter is a processor that determines if the data structure satisfies
some criteria but does not modify it.
- A pruner removes unwanted banks before output so does modify the data
structure, but far from adding value, it subtracts it!
- The Monte Carlo and the calibration produce the same outputs. The same
is true for the uncalibration and the unpacker. Processors that produce the
same outputs cannot interwork asynchronously; whichever is the downstream one
permantly modifies the results of the upstream one.
So, in drawing up rules to allow processors to be used as independently as
possible from one another, we have to recognise that not all can be
asynchronous. Instead we have one set of rules which, if observed, qualifies
the processor to be called a Standard Processor, and an additional set of
rules govern the operation of any Asynchronous Processor. In SNOMAN, all
processors will be Standard Processors, but not all will be Asynchronous
Processors. Indeed currently, only the fitters are asynchronous. More work
needs to be done in this area when we assemble an interactive version of
SNOMAN.
Next: Processors: Definitions and Rules
Up: Processors: A Discussion
Previous: Maintaining Data Structure Integrity
Contents
sno Guest Acct
2009-09-09