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
When an extractor fails and the verbose flag is on it prints a custom error message with the classification ID cause issues, the reducer should do the same thing. This message should include the subject ID that is failing to reduce and possible the classification ID of all extracts (if those are easy to pull out).
This will aid in troubleshooting when things go wrong.
In addition to this new verbose flag, the documentation should be updated with a "troubleshooting" section that explains how to turn on these flags and how to interpret what the message is saying, how to report the issue, or how to "skip" over the issue (i.e. delete a row for the CSV just to get the aggregation to run if the loss of one classification is acceptable).
Example of a time when the loss of classification is acceptable could be if a classification from the wrong workflow is included in the data set (this was an old Panoptes bug that is not an issue anymore, but it is good to have an example).
The text was updated successfully, but these errors were encountered:
When an extractor fails and the verbose flag is on it prints a custom error message with the classification ID cause issues, the reducer should do the same thing. This message should include the subject ID that is failing to reduce and possible the classification ID of all extracts (if those are easy to pull out).
This will aid in troubleshooting when things go wrong.
In addition to this new verbose flag, the documentation should be updated with a "troubleshooting" section that explains how to turn on these flags and how to interpret what the message is saying, how to report the issue, or how to "skip" over the issue (i.e. delete a row for the CSV just to get the aggregation to run if the loss of one classification is acceptable).
Example of a time when the loss of classification is acceptable could be if a classification from the wrong workflow is included in the data set (this was an old Panoptes bug that is not an issue anymore, but it is good to have an example).
The text was updated successfully, but these errors were encountered: