Delete custom dataset

[boriskovar-m2ms]

User will be able to delete results of the jobs which are computed dataset which are displayed in the right hand side.

API call to delete given dataset: DELETE api/compound-sets/

User will be able to delete the sets from jobs table. User will be able to select multiple job results and press the button Remove. The system will show a confirmation dialog with text "Are you sure you want to delete following datasets: " with two buttons Yes and No. After confirmation (Yes button pressed) system will remove tab for each selected computed dataset from RHS and after that call DELETE api/compound-sets/ for each selected job. Pressing the No will cancel the action.

[ag-m2ms]

This might be a bigger problem than what it looks like. There is a handful of edge cases which have to be handled:

• Molecules of a dataset which is being deleted might be displayed in the NGL viewer. The correct behavior is to remove them from the viewer.
• Actions related to the dataset are tracked for undo/redo operations. What should happen upon deletion is to remove all actions related to the dataset.
• A snapshot with actions related to the dataset might have been created. In this scenario, when loading up the snapshot the actions regarding the deleted dataset should be skipped. This should already be present according to @boriskovar-m2ms (needs to be verified).
• The deleted dataset might be selected in the RHS. If that's the case the selected dataset should reset and the default first tab will be displayed.
• Molecules of the dataset could be selected. They should be unselected.
• Updated to the dataset's filter is tracked as well. Not sure if this should be cleared since once the dataset is gone, there is basically no way of accessing it.
• Using up and down arrows in RHS. The second point applies.
• Tab switching on RHS. This one is a tricky one since this action doesn't involve storing datasetID in dataset_id/datasetID. It stores its title in object_id field. For one I am not sure if title is unique, secondly object_id is for general purposes only. I could add a new field into the action, it might break existing snapshots however.
• The library used for undo and redo doesn't seem to support history replacing, only clearing. I assume the best thing we can do is probably just to ignore the actions.

If I encounter any other edge cases I will update this comment.

[ag-m2ms]

This tracks updates to the points mentioned above:

• ~Molecules of a dataset which is being deleted might be displayed in the NGL viewer. The correct behavior is to remove them from the viewer.~
• Actions related to the dataset are tracked for undo/redo operations. What should happen upon deletion is to remove all actions related to the dataset.
• A snapshot with actions related to the dataset might have been created. In this scenario, when loading up the snapshot the actions regarding the deleted dataset should be skipped. This should already be present according to @boriskovar-m2ms (needs to be verified).
• ~The deleted dataset might be selected in the RHS. If that's the case the selected dataset should reset and the default first tab will be displayed.~
• ~Molecules of the dataset could be selected. They should be unselected.~
• Updated to the dataset's filter is tracked as well. Not sure if this should be cleared since once the dataset is gone, there is basically no way of accessing it.
• ~Using up and down arrows in RHS. The second point applies.~
• Tab switching on RHS. This one is a tricky one since this action doesn't involve storing datasetID in dataset_id/datasetID. It stores its title in object_id field. For one I am not sure if title is unique, secondly object_id is for general purposes only. I could add a new field into the action, it might break existing snapshots however.
• The library used for undo and redo doesn't seem to support history replacing, only clearing. I assume the best thing we can do is probably just to ignore the actions.

[boriskovar-m2ms]

@rsanchezgarc @phraenquex

Results from the jobs will be uploaded to the fragalysis and user will able to see and inspect these results directly in the fragalysis front-end. Every action is stored as an action so user can undo-redo these actions, create a project (which will save a snapshot), save an anonymous snapshot so someone else can restore the snapshot which in turn will replay last action (if any) on given objects. It’s important to understand that it doesn’t matter if the dataset is only marked as deleted and not physically deleted on the backend, but if frontend doesn’t receive any information about existence of the dataset then, from frontend point of view, it doesn’t matter if it is physically or only logically deleted. There are two basic modus operandi for a normal user.

1. Free form work – In this mode user doesn’t work within framework of the projects. User can perform actions on the dataset and after that delete given dataset. a. User refreshes the page – actions are lost, and everything is fine b. User doesn’t refresh the page – i. What to do when he tries to undo/redo actions? Disable the buttons and tell the user somewhere to reload the page? Try to just ignore actions which are performed on the missing objects and notify user that given actions was skipped (Do we need to indicate this also on the actions timeline)? ii. What to do when user tries to save a snapshot? Disable this functionality until user does refresh the page? Or let user save and if object doesn’t exist action on that object is not stored. If snapshot already exists, then actions on no longer existing objects are ignored (do we need to notify user?).
2. Working on the project – Here it is very important to understand, compared to free form work, there is an autosave feature which after five actions stores these actions in the backend so the invalid actions are already there, so I THINK we have no other option then just to make sure that actions on invalid objects are ignored.

Ignoring of the actions performed on the missing objects when loading snapshot should be present but should be verified and fixed where it's not so.