-
Notifications
You must be signed in to change notification settings - Fork 2
Long Running Replications
Michael Darisse edited this page Sep 11, 2020
·
1 revision
In general, if there seems to be long-running code, either from the last time we saw a paper, or because there is a note in the README (specifying the time requirement),
-
leave it running on CISER for a few days. You can work on other issues in the meantime (but be cognizant of CISER downtimes)
- Post a note to JIRA comment that code is still running
-
If after a few days, the code has not finished, write up the report as an "incomplete replication", and submit the report
- Make it in the report that there is still code running.
- Leave the code running as long as possible!
- Try to get all other code (if independent) to run as well.
- One of the reasons for incomplete reproduction is then ‘insufficient time’ (one of the checkboxes on Jira).
-
What to do with the running code?
- often, this happens during the first submission of a paper. We will request that authors specify if this is a bug, or if there is a LONG time requirement.
- When the paper comes back, we may have the result of the long-running program, so keep updating the repository for these issues with log files, output, etc. and make a note in the JIRA issue if the code finishes running.
-
Training
-
Tips for authors
-
Tips for replicators
-
Questionnaires
-
Definitions
-
Generic workflow
-
Post-publication replications
-
Technical issues
-
Appendix