Page MenuHomePhabricator

Prepare tests to cover missing SQLiteQueryExecutor queries
ClosedPublic

Authored by marcin on Jan 18 2024, 9:21 AM.
Tags
None
Referenced Files
Unknown Object (File)
Sat, Apr 13, 10:28 PM
Unknown Object (File)
Mar 30 2024, 12:44 AM
Unknown Object (File)
Mar 28 2024, 1:29 PM
Unknown Object (File)
Mar 7 2024, 10:11 AM
Unknown Object (File)
Mar 7 2024, 6:52 AM
Unknown Object (File)
Feb 24 2024, 10:22 PM
Unknown Object (File)
Feb 20 2024, 9:55 PM
Unknown Object (File)
Feb 19 2024, 2:08 PM
Subscribers
Tokens
"Love" token, awarded by kamil.

Details

Summary

I created it in order to provide unit tests that can serve as a test plan for previous differentials in this stack since most of them break the app.

Test Plan

Execute implemented tests.

Diff Detail

Repository
rCOMM Comm
Lint
Lint Not Applicable
Unit
Tests Not Applicable

Event Timeline

Add unit tests for Messages and Media tables

native/cpp/CommonCpp/DatabaseManagers/entities/Message.h
82 ↗(On Diff #35895)

I had issues with passing int and std::unique_ptr<int> between JS and C++ so I decided to give up since Messages queries don't use integer fields in Message structure so mocking them with some defaults doesn't affect credibility of tests.

If we ever consider to land this differential we can workaround by converting int <-> string

Is there a reason we can't land the unit tests?

Implement tests for olm data tables and queries.

Is there a reason we can't land the unit tests?

Initially I didn't plan to do so but I don't see a reason we shouldn't. One thing to note is that if we want to land unit tests we need a closer review from @kamil since this diff introduces some work he would need to do in future when migrating the rest of redux to SQLite on web so we need to ensure this work is done in a way he would expect to.

kamil requested changes to this revision.Jan 25 2024, 9:41 AM

Overall this diff is super helpful, adds tests, and introduces code that will be needed in the future anyway.

Three things:

  • Please use camelCase on the JS side.
  • remove those default values on WebMessage
  • split storeOlmPersistDataWeb into two functions - it will be easier to adapt ops on the web later with this code
native/cpp/CommonCpp/DatabaseManagers/DatabaseQueryExecutor.h
19–22 ↗(On Diff #35901)

Maybe it's better to define this in Message.h?

98 ↗(On Diff #35901)

I would prefer to fo split this into two separate functions (on the web), one for persisting account and one for persisting session. I think when only the session is updated we can avoid persisting the same account again.

native/cpp/CommonCpp/DatabaseManagers/entities/Message.h
99 ↗(On Diff #35901)

why this is 0?

web/cpp/SQLiteQueryExecutorBindings.cpp
66 ↗(On Diff #35901)
web/database/types/sqlite-query-executor.js
11 ↗(On Diff #35901)
27 ↗(On Diff #35901)
49–51 ↗(On Diff #35901)

we should user camel case in JS

This revision now requires changes to proceed.Jan 25 2024, 9:41 AM
  1. Update to camel case in JS
  2. Move MessageWithMedias type to Message.h
  3. Split olm account and session persistence.
  4. Make WebMessage type complete.
kamil requested changes to this revision.Feb 2 2024, 4:30 AM

A couple of things but it's very close

native/cpp/CommonCpp/DatabaseManagers/DatabaseQueryExecutor.h
80–82 ↗(On Diff #36357)

to make it useful on the web I think we should also add the id of both account and session (cc. @michal who is working on moving crypto things to worker on web)

native/cpp/CommonCpp/DatabaseManagers/SQLiteQueryExecutor.cpp
1892–1893 ↗(On Diff #36357)

Can't find where it was introduced but I think it can be deleted now? (because of D10703)

web/cpp/SQLiteQueryExecutorBindings.cpp
31 ↗(On Diff #36357)

can you in one of the stores add a test that checks this value to make sure everything works?

163–174 ↗(On Diff #36357)

I think you can put it above to group it logically

web/database/_generated/comm-query-executor.js
119 ↗(On Diff #36357)
web/database/queries/messages-and-media-queries.test.js
156–158 ↗(On Diff #36357)

you can use find

web/database/queries/olm-persist-data-queries.test.js
40 ↗(On Diff #36357)

I think it's safer to check value

web/database/queries/threads-queries.test.js
78 ↗(On Diff #36357)

message store threads is something different

This revision now requires changes to proceed.Feb 2 2024, 4:30 AM
native/cpp/CommonCpp/DatabaseManagers/DatabaseQueryExecutor.h
80–82 ↗(On Diff #36357)

We can't do it now unfortunately since the implementation of getOlmPersistAccountData will throw an error if if fetches more than two rows it will throw an error. We will be able to do it after https://linear.app/comm/issue/ENG-6510/persist-notifications-crypto-account-in-sqlite-database is done.

native/cpp/CommonCpp/DatabaseManagers/DatabaseQueryExecutor.h
80–82 ↗(On Diff #36357)

We should make a note of this on Linear somewhere so it's not missed later

native/cpp/CommonCpp/DatabaseManagers/DatabaseQueryExecutor.h
80–82 ↗(On Diff #36357)

I think the task I linked is enough. As a part of this task we will introduce id argument to the function and compilation will force us to update web code as well.

  1. Use find in tests
  2. Test for isNull correctness for NullableInt
  3. Fix naming in ThreadStore tests.
  4. Check for value of NullableString in OlmStore tests
  5. Reorder olm-related method definitions.

Good job, this diff is super helpful

web/database/queries/olm-persist-data-queries.test.js
8 ↗(On Diff #36605)
This revision is now accepted and ready to land.Feb 5 2024, 1:51 PM

Fix olm tests description