Summary
The class masking method's "Source classifier" picker lists classification algorithms on the project's currently enabled pipelines. A classifier that produced this project's existing results through a pipeline the project has since switched off, or one derived by an earlier masking run, is missing from it, although the task can process its predictions. The API accepts any existing algorithm, so this only affects the dialog.
The picker should list the classifiers that actually produced results in the chosen capture set, as the dialog design asks ("only classifiers that ran on this set are listed"). occurrences/algorithms already lists a project's historical algorithms and would need a task_type filter (and ideally a capture-set filter). Measure the query on a large project before switching, since filters over classifications scale with table size.
List of Changes
- The class masking picker offers every classifier whose results are in the chosen capture set.
Raised in review of #1447.
🤖 Generated with Claude Code
Summary
The class masking method's "Source classifier" picker lists classification algorithms on the project's currently enabled pipelines. A classifier that produced this project's existing results through a pipeline the project has since switched off, or one derived by an earlier masking run, is missing from it, although the task can process its predictions. The API accepts any existing algorithm, so this only affects the dialog.
The picker should list the classifiers that actually produced results in the chosen capture set, as the dialog design asks ("only classifiers that ran on this set are listed").
occurrences/algorithmsalready lists a project's historical algorithms and would need atask_typefilter (and ideally a capture-set filter). Measure the query on a large project before switching, since filters over classifications scale with table size.List of Changes
Raised in review of #1447.
🤖 Generated with Claude Code