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
the last time that i checked, hpcviewer used a sparse set of (sometimes) large numbers for metric indices. it should not present such numbers. if such numbers arise (perhaps while computing metrics in the viewer), we should build a bi-directional map from internal sparse metric indices to a set of dense small indices. a hash table can be used for the sparse to dense mapping. an array can map back to internal sparse metric indices if necessary.
The text was updated successfully, but these errors were encountered:
To support backward compatibility, it's much simpler to renumbering the
metric index inside the derived metric window instead of in the
beginning.
If we renumber in the beginning, we'll have a headache with storing the
formula as the metric index will differ from experiment.xml to the
stored one.
This may not be an issue in prof2 since we don't have metric id. But
since we need to keep backward compatibility, we should make changes
with minimal effects.
* Code clean up, remove dead codes
* Fix issue #168 (incorrect metric when merging stat metrics)
* Fix build.yml to include develop branch
* Generalize database parser for v2 and v4
Add DataMeta to parse database v4 (from prof).
* Change the API to open a database: use File instead of String
* Rename maven groupId to edu.rice.cs.hpcviewer from e.r.hpcviewer
* refine unit test for Experiment class
* print the error message instead of the stack trace
* Fix security hotspot and other issues reported by sonar
* Standardize the comments
* Derived metric window with content assist
* Draft fix issue #151 (renumbering metric index)
To support backward compatibility, it's much simpler to renumbering the
metric index inside the derived metric window instead of in the
beginning.
If we renumber in the beginning, we'll have a headache with storing the
formula as the metric index will differ from experiment.xml to the
stored one.
This may not be an issue in prof2 since we don't have metric id. But
since we need to keep backward compatibility, we should make changes
with minimal effects.
* Refine the test
* Minor Fixes to conform with Java compliant
the last time that i checked, hpcviewer used a sparse set of (sometimes) large numbers for metric indices. it should not present such numbers. if such numbers arise (perhaps while computing metrics in the viewer), we should build a bi-directional map from internal sparse metric indices to a set of dense small indices. a hash table can be used for the sparse to dense mapping. an array can map back to internal sparse metric indices if necessary.
The text was updated successfully, but these errors were encountered: