pdr-backend
                                
                                 pdr-backend copied to clipboard
                                
                                    pdr-backend copied to clipboard
                            
                            
                            
                        [Lake][Subgraph][App] Update app/ & drop: PredictSlot and subgraph. Use Slot & lake.
Background / motivation
There are various functions inside app + subgraph, that are queryng directly from subgraph + are using an object/data schema called PredictSlot.
This was causing some conflicts w/ lake, and the object that we're using there: "Slot".
Generally, we want to be building functionality that queries from the lake... and wherever that's not possible, to query from an API (that is yet to be built).
In the current cases:
- pdr-backend/app
Is still using PredictSlot, and it looks like all hooks/information, is there to mainly support app/ which to-this-day, still powers the predictoor.ai frontend.
TODOs / DoD
- Reconcile these two things so we have 1 object: Slot
- Move everything to query/get data from the lake (or a remote API)
- Port/Update all the work inside of app/elsewhere that's using Subgraph + Predict Slot => To use Lake + Slot
- Remove/Drop all usage of PredictSlot
- Have all exterior facing systems querying from the lake
Tasks:
- [x] Stop using PredictSlot and drop it from code
- [x] Stop querying the subgraph for answers related to Slots, use the lake instead
- [x] Update APP to query from the lake's "Slot" table, to provide information
- [x] Update other APIs in app/to serve the correct information from the lake
- [ ] Write README.md on how to deploy server, update lake, and leave lake updating
- [ ] Write README.md for how to test the server, to make sure that it's working and updating correctly
- [ ] Make sure that predictoor.ai FE is still working using latest implementation that's serving from lake
Please close when https://github.com/oceanprotocol/pdr-backend/issues/1048 is complete
@idiom-bytes IMO we should not keep 2 tickets if only #1048 is relevant. It crowds the workspace. If the pending TODOs have been abstracted into #1048, I believe we can close this.
Tracking outstanding items (including this) in #1299 and closing this to cleanup the backlog, until we have it resolved.