Project

General

Profile

Feature #21715

Feature #21229: 'Shuffling' input module

Determine product lookup policy for sampled products

Added by Kyle Knoepfel 10 months ago. Updated 10 months ago.

Status:
Closed
Priority:
Normal
Assignee:
Category:
Infrastructure
Target version:
Start date:
01/17/2019
Due date:
% Done:

100%

Estimated time:
Spent time:
Scope:
Internal
Experiment:
Mu2e
SSI Package:
art
Duration:

Description

The SamplingInput source is responsible for creating the Sampled<T> (sub)run products, where T is the specific product type. As the products are created by the input source, a sensible means of retrieving the products from the input datasets needs to be developed. Specifically, the original module label, instance name, and process name must, to some extent, be preserved in order to disambiguate between the products from the input datasets.

History

#1 Updated by Kyle Knoepfel 10 months ago

  • % Done changed from 0 to 100
  • Status changed from New to Resolved

For a job that uses the SamplingInput source, the chosen behavior is:

  • Event data products from input files are retrievable using the same behavior as listed here.
  • No (sub)run products from input files are directly retrievable, nor are they persisted to an output file. Instead, (sub)run products are accessed through the art::Sampled<T> class template, which can be persisted to disk.
  • For looking up Sampled<T> products, users specify the original module label and instance name for the requested product. However, instead of specifying the original process name (e.g. "MakeTracks"), the "SampledFrom" prefix is prepended to the original process name (e.g. "SampledFromMakeTracks"). Users are encouraged to call art::sampled_from("MakeTracks"), as the process name, which will return the correctly prefixed process name.

Please see here for full documentation.

#2 Updated by Kyle Knoepfel 10 months ago

  • Status changed from Resolved to Closed


Also available in: Atom PDF