Repository navigation
Edit (some) fields of flow after upload? #896
Description
I noticed I have some uploaded flows for which I would like to add a custom_name. I'm currently unable to do so through the Python API (or at all). This gives rise to the following questions:
- Should we support filling in the
custom_namefield later if left empty initially? - Are there other attributes for which this makes sense?
- Should we support editing
custom_name(or other fields) after uploading? - Should we look at other entities as well to see what we might want to be editable?
Of course it is important to preserve consistency. Especially if other users start using your flow, you don't want it to suddenly represent something else.
Personally I feel that providing a custom_name after not having one at upload time is fine. I would say editing custom_name is OK too, considering we auto-generate them with the scikit-learn extension for some flows (and if that goes wrong, editing is probably preferred?).
Other fields that seem candidates for editing to me are: tags, language and binary_url (probably not md5/format).
I'm not sure to which extend this is a design issue or just openml-python.
Activity
joaquinvanschoren commentedon Jun 22, 2020
This requires a new REST API endpoint.
We then need to discuss for which fields we want to allow editing.
For inspiration, see the data edit pull request by Sahi:
openml/OpenML#1044
PGijsbers commentedon Jul 23, 2020
PR related to that data edit change for openml-python: #935
mfeurer commentedon Jun 15, 2021
Hey, is this issue resolved? The two PRs above are both merged
PGijsbers commentedon Jul 2, 2021
No. The changes were only made to allow edits to dataset entities. OpenML Python still does not support changing the flows as far as I can tell (nor does the Rest API, I think).
feat: Add edit_flow function to allow editing of flow metadata
Prachi194agrawal commentedon Dec 9, 2025
Hello everyone,
I've opened Pull Request #1527 (linked above) to address this issue by implementing the client-side functionality for editing flow metadata.
The new function,
openml.flows.edit_flow, allows authenticated users to update the following non-critical fields on an existing flow:custom_nametagslanguagedescriptionThis implementation relies on the existence of the corresponding
flow/editREST API endpoint.I believe this completes the work needed on the
openml-pythonside to close Issue #896. Could a maintainer please review PR #1527?Thanks!
feat: Implement openml.flows.edit_flow for metadata editing (Closes o…