CR-215 workbench local storage history

[pd-redis]

In this PR, I am checking if envDependent flag is off, and if it is, do not perform API request to list/get/delete Workbench commands.
Also, if the flag is off, I am storing results of sendWBCommandAction and sendWBCommandClusterAction in localStorage, from where they can be loaded after

[rsergeenko]

IMO you should not use localStorage to save history of workbench, better to use IndexedDB API. LocalStorage has limited size of memory, also we limit the size of history results (only 20 results per database)

[rsergeenko]

Please, also be sure that users can work with different databases and see results per database

[pd-redis]

IMO you should not use localStorage to save history of workbench, better to use IndexedDB API. LocalStorage has limited size of memory, also we limit the size of history results (only 20 results per database)

I know that localStorage is limited, but still fas few MB size (about 5). Do you think it is reasonable to expect that 20 commands can exceed that?

[rsergeenko]

IMO you should not use localStorage to save history of workbench, better to use IndexedDB API. LocalStorage has limited size of memory, also we limit the size of history results (only 20 results per database)

I know that localStorage is limited, but still fas few MB size (about 5). Do you think it is reasonable to expect that 20 commands can exceed that?

yes, even 1 command result can exceed localStorage size. Users may have several databases, in this case they will have (number of databases * 20) results. Also 1 command can be in group mode and contain several results of commands.

[pd-redis]

IMO you should not use localStorage to save history of workbench, better to use IndexedDB API. LocalStorage has limited size of memory, also we limit the size of history results (only 20 results per database)

I know that localStorage is limited, but still fas few MB size (about 5). Do you think it is reasonable to expect that 20 commands can exceed that?

yes, even 1 command result can exceed localStorage size. Users may have several databases, in this case they will have (number of databases * 20) results. Also 1 command can be in group mode and contain several results of commands.

OK, I will look into it

[pd-redis]

IMO you should not use localStorage to save history of workbench, better to use IndexedDB API. LocalStorage has limited size of memory, also we limit the size of history results (only 20 results per database)

I know that localStorage is limited, but still fas few MB size (about 5). Do you think it is reasonable to expect that 20 commands can exceed that?

yes, even 1 command result can exceed localStorage size. Users may have several databases, in this case they will have (number of databases * 20) results. Also 1 command can be in group mode and contain several results of commands.

Implemented indexedDb storing

[rsergeenko]

I think its better to move class IndexedDbStorage to a different file. Also, I would recommend to have a file with instances of different tables (export const wbHistoryStorage = new IndexedDbStorage('RI_WB_HISTORY'))

IMO it's better to have a separate record for each command execution (like implemented on BE side) than modify each time one.

Will limit of records for 1 database be implemented?

[pd-redis]

I think its better to move class IndexedDbStorage to a different file. Also, I would recommend to have a file with instances of different tables (export const wbHistoryStorage = new IndexedDbStorage('RI_WB_HISTORY'))

Sure, moving it to separate file makse sense. I don't think we'll have other databases besides wb history one, at least for now. Which is why, I didn't bring external library to manage this.

IMO it's better to have a separate record for each command execution (like implemented on BE side) than modify each time one.

You mean like, every response from POST /workbench/command-executions to be separate record?

Will limit of records for 1 database be implemented?

yeah, this is stil Work In Progress

[rsergeenko]

You mean like, every response from POST /workbench/command-executions to be separate record?

yes, I think you can try to use indexes https://developer.mozilla.org/en-US/docs/Web/API/IDBObjectStore/index

[rsergeenko]

@pd-redis What will be if we need to add one more table/storage?

I think it will be better to use the library (https://dexie.org/) to be more flexible and stable.
Also, I think we should discuss how to work with local database and sql (maybe we can implement one solution for all build types?)
@ArtemHoruzhenko @zalenskiSofteq - what do you think?

[seth-riedel]

Please see comments. I think we also need a lot of tests for this, both for the changes to the workbench slice as well as the behavior of services/workbenchStorage.ts

[pd-redis]

@pd-redis What will be if we need to add one more table/storage?

I think it will be better to use the library (https://dexie.org/) to be more flexible and stable. Also, I think we should discuss how to work with local database and sql (maybe we can implement one solution for all build types?) @ArtemHoruzhenko @zalenskiSofteq - what do you think?

Reg dexie - we discussed it, but the consensus was, that bringing 90K dependency just to store WB history is not worth it.
Otherwise, I have used it before and I like it.

[ArtemHoruzhenko]

@pd-redis will this approach support storing other data in the future? Are we ready to add another collection to idb without migrating existing data?