This code will switch redux-persist storage to this based on encrypted SQLite.
However, to make a gentle switch we need to look to old storage and migrate the data if possible.
Depends on D7667
Details
Details
- Enable this code
- Check is persisted content is not missing but stored in SQLite
- Data from old storage should be removed.
Diff Detail
Diff Detail
- Repository
- rCOMM Comm
- Lint
Lint Not Applicable - Unit
Tests Not Applicable
Event Timeline
| web/redux/create-async-migrate.js | ||
|---|---|---|
| 28–46 ↗ | (On Diff #25854) | Is it a standard way of doing this? Wondering if checking this during each migration makes sense. Is it possible that we would want to move from one store to the other without any new migration introduced? |
| web/redux/create-async-migrate.js | ||
|---|---|---|
| 28–46 ↗ | (On Diff #25854) |
I didn't find any suggestion of how to do it properly, there is one issue on GitHub and my solution is similar to the comments there (link).
We not checking in each migration, we check once, one app start, and rehydrated state is empty.
Yes, this code will migrate storage engines without any new migrations. I can explain two alternative solutions but I don't think they're better:
According to this comment: https://phab.comm.dev/D7658?id=25832#inline-50059. Run through migrations and if the undefined state was an argument, return undefined, and then introduce the next migration which will be responsible for taking a look into the old storage engine.
This callback is called after rehydration and migrations, we can look into the old store but it's tough to propagate data to a new one. Let me know what do you think. |
Comment Actions
I tried a different approach (getting old storage inside migration) but it will not be better. We need to have a store version before the migration process, and in that case, this version is persisted via old storage, so we need to retrieve it here.
| web/redux/create-async-migrate.js | ||
|---|---|---|
| 45 ↗ | (On Diff #26355) | Is it possible that something goes wrong (e.g. an exception is thrown) and we remove old state before saving it to a new place? Can we purge only after the successful migration to SQLite? |
| web/redux/create-async-migrate.js | ||
|---|---|---|
| 45 ↗ | (On Diff #26355) | In this case, I don't consider this as an issue. This code is responsible only for reading persisted data into memory (and changing it if there are some migrations) and then with the next modifications data will be persisted in the new engine (that being said I don't think we should leave this data here, and we can purge this). Problems with saving data into new storage are separate things which mean something is wrong with the database and we should fetch new data for the user - in this case, it'll be reloaded. |