You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In the future, outputs of the statistics classes would be done with PackCollection & TimeHistory & 2 events (3 in total), which is quite verbose and inconvenient for the user to maintain.
Proposed cleanup
I propose 2 alternatives but these suggestions could change:
Main Proposal:
We could only have 2 node to ask When and What is to output.
CompositionalMultiphaseStatistics would now be an output, with two optional attributes:
logLevel to enable the reporting in the log, necessary for the user to quickly see the simulation state,
outputFile (s?), to allow the output of those statistics in a file (HDF5, CSV...).
Within the code, we would have:
To gather the data, CompositionalMultiphaseStatistics (and log it if requested),
To output the data, the FieldStatisticsBase (a new StatisticsOutput class) which would create automatically the necessary TimeHistoryOutput, PackCollection and their events.
Separate the output file from the statistics declaration, therefore we can have multiple statistics output files inside <Outputs>. StatsOutput would be an observer of the statistics classes and wouldn't require additional events.
Describe the issue
In the future, outputs of the statistics classes would be done with
PackCollection&TimeHistory& 2 events (3 in total), which is quite verbose and inconvenient for the user to maintain.Proposed cleanup
I propose 2 alternatives but these suggestions could change:
Main Proposal:
We could only have 2 node to ask When and What is to output.
CompositionalMultiphaseStatisticswould now be an output, with two optional attributes:logLevelto enable the reporting in the log, necessary for the user to quickly see the simulation state,outputFile(s?), to allow the output of those statistics in a file (HDF5,CSV...).Within the code, we would have:
CompositionalMultiphaseStatistics(and log it if requested),CompositionalMultiphaseStatistics::RegionStatisticsGroup, as proposed by @untereiner in Collect log information inhdf5#2298,FieldStatisticsBase(a newStatisticsOutputclass) which would create automatically the necessaryTimeHistoryOutput,PackCollectionand their events.Alternative Proposal
Separate the output file from the statistics declaration, therefore we can have multiple statistics output files inside
<Outputs>.StatsOutputwould be an observer of the statistics classes and wouldn't require additional events.co-authored with @MelReyCG.