We need to think through sharing / linking across projects and how it interacts with publishing / unpublishing datasets. I think we should encourage linking to objects in published datasets, especially if they have a doi, discourage changing / unpublishing them, and enable some ways of cloning/versioning datasets; adding to a dataset without modifying existing elements; or warning about unpublishing if there are downstream links.
The current process is fairly likely to be overwhelming for a project with a CASM-like number of objects if you want to not just include the entire experiment. I'm not quite sure how to deal with that, but getting dataset creation into the CLI could help. Maybe some faceted-search type filters (say on process type, regex matches against sample or process name, attribute existance, attribute value range, etc.) will be necessary. This is another place where I think "selection" creation and editing could be useful since you could do a query to create a selection and then add/remove those objects. Also, support for moving objects between experiments could help some.
longer term / next iteration:
We need to think through sharing / linking across projects and how it interacts with publishing / unpublishing datasets. I think we should encourage linking to objects in published datasets, especially if they have a doi, discourage changing / unpublishing them, and enable some ways of cloning/versioning datasets; adding to a dataset without modifying existing elements; or warning about unpublishing if there are downstream links.
The current process is fairly likely to be overwhelming for a project with a CASM-like number of objects if you want to not just include the entire experiment. I'm not quite sure how to deal with that, but getting dataset creation into the CLI could help. Maybe some faceted-search type filters (say on process type, regex matches against sample or process name, attribute existance, attribute value range, etc.) will be necessary. This is another place where I think "selection" creation and editing could be useful since you could do a query to create a selection and then add/remove those objects. Also, support for moving objects between experiments could help some.