Describe the bug
When extracting a workflow from a history that includes a tool step which consumed User-Defined Tool (UDT) inputs, the extracted workflow is incomplete. The intermediate tool step that used the UDTs is dropped from the resulting workflow, leaving the workflow disconnected and non-runnable.
Galaxy Version and/or server at which you observed the bug
Galaxy Version: test.galaxyproject.org (Test server)
Server: https://test.galaxyproject.org
To Reproduce
Steps to reproduce the behavior:
- Use a history that runs an analysis where one tool step takes two UDTs (User-Defined Tools / data sources) as inputs.
- Extract a workflow from this history (Histories → Extract Workflow).
- Open the extracted workflow / view the workflow preview.
Expected behavior
The extracted workflow should contain both tool steps with all their inputs preserved:
- Step 1: input dataset
pangenome4.fa
- Step 2: PGGB Pangenome Aligner (consuming
pangenome4.fa plus its two UDT inputs)
- Step 3: Synteny strip plot from PGGB graph (consuming the output of PGGB)
All connections between steps should be intact and the workflow should be runnable.
Actual behavior
The extracted workflow only contains two nodes:
1: pangenome4.fa (input)
2: Synteny strip plot from PGGB graph (tool)
The PGGB Pangenome Aligner step is completely missing. The graph_gfa input port on the Synteny step is unconnected (shown with a red asterisk indicating a required input is missing). The UDT inputs are also not represented anywhere in the workflow.
Extracted workflow URL (for reference): https://test.galaxyproject.org/published/workflow?id=cbed8cea36b844a7
Screenshots
(Screenshots of the source history and the resulting incomplete workflow canvas are available and can be attached on request.)
Additional context
This appears to be related to how the workflow extraction logic handles tool steps whose inputs come from UDTs — those steps seem to be silently dropped instead of being included (or at least flagged) in the extracted workflow. As a result, the extracted workflow is not a faithful representation of the analysis performed in the history.
Describe the bug
When extracting a workflow from a history that includes a tool step which consumed User-Defined Tool (UDT) inputs, the extracted workflow is incomplete. The intermediate tool step that used the UDTs is dropped from the resulting workflow, leaving the workflow disconnected and non-runnable.
Galaxy Version and/or server at which you observed the bug
Galaxy Version: test.galaxyproject.org (Test server)
Server: https://test.galaxyproject.org
To Reproduce
Steps to reproduce the behavior:
pangenome4.fa(input fasta)PGGB Pangenome Aligner on dataset 1(output of the PGGB tool which consumed the UDTs)Synteny strip plot from PGGB graph on dataset 2(output of the Synteny tool)Expected behavior
The extracted workflow should contain both tool steps with all their inputs preserved:
pangenome4.fapangenome4.faplus its two UDT inputs)All connections between steps should be intact and the workflow should be runnable.
Actual behavior
The extracted workflow only contains two nodes:
1: pangenome4.fa(input)2: Synteny strip plot from PGGB graph(tool)The PGGB Pangenome Aligner step is completely missing. The
graph_gfainput port on the Synteny step is unconnected (shown with a red asterisk indicating a required input is missing). The UDT inputs are also not represented anywhere in the workflow.Extracted workflow URL (for reference): https://test.galaxyproject.org/published/workflow?id=cbed8cea36b844a7
Screenshots
(Screenshots of the source history and the resulting incomplete workflow canvas are available and can be attached on request.)
Additional context
This appears to be related to how the workflow extraction logic handles tool steps whose inputs come from UDTs — those steps seem to be silently dropped instead of being included (or at least flagged) in the extracted workflow. As a result, the extracted workflow is not a faithful representation of the analysis performed in the history.