Differential D11481
[sqlite/lib] Split keyserver ops for backed and non-backed up data Authored by michal on Mar 29 2024, 1:43 PM.
Details ENG-6928 : Return the default data after restoration Depends on D11473 Run the keyservers-queries.test.js. Make sure that keyserver tables are filled in with data correctly. Check that if the non-synced data doesn't have any data, the ops return default values.
Diff Detail
Event TimelineComment Actions Accepting but I am of strong opinion that it would be better to create two new structs: NonSyncedKeyserverInfo and SyncedKeyserverInfo that would look exactly the same as original KeyserverInfo. In SQLiteQueryExecutor methods we would just create instances of each class and could rely on simple SQL queries without named parameters. The code would be easier to understand and maintain. It would be two small classes doing simple logic instead of one bigger doing complex logic. Comment Actions Agree this breaks our convention a bit but not sure if this is more maintainable - I think it's better to have one entity which is some sort of shared interface on how both C++ and JS understand data, the only difference is that we have two tables for it, but on each operation, we modify both and when querying we joined them. I also think named parameters are better overall, we should implement them that way, now probably there will be no time to prioritize them.
Comment Actions Small fixes I agree with @kamil, these two tables are bundled together logically and the operations should be made on both of them, so splitting ops in two would make it easier to introduce bugs and subtle differences. | ||||||||||||||||||||||