Update to typeorm 0.3.x

[javiercr]

Context

• Slack thread 1
• Slack thread 2
• typeorm changelog

We've identified that often the cause of the "Waiting for this message" problem is an error when decrypting the message node and calling either to repository.decryptGroupMessage or repository.decryptMessage.

The actual error in this scenario is The database connection is not open.

Both decryptGroupMessage and decryptMessage need to call the signal storage:

• For group messages, it needs to load the senderKey from the storage
• For regular messages, it needs to load the session from the storage.

The signal storage we're providing to Baileys is a backed by a typeorm sqlite database connection, therefore the actual error (The database connection is not open) comes from platform-whatsapp -> typeorm -> better-sqlite3.

Description

We still don't know the actual cause of the database connection not being initialized when baileys / libsignal need to talk to it. However typeorm 0.3.0 replaced the Connection abstraction with a new DataSource class.

This PR bumps typeorm from 0.2.41 to 0.3.17 with the hope that some of the latest changes in typeorm could help mitigate the db connection problem.

Even if this doesn't solve the problem described, given the amount of breaking changes in typeorm from 0.2.x to 0.3.x I do think it's worth it updating now to avoid problems in the future.

Summary of relevant changes:

• Rename Connection to DataSource
• Rename findOne({ id }) to findOneBy({ id }) and findOneOrFail({ id }) to findOneByOrFail({ id })
• Add missing where key to find() args
• Use findOne with a (mandatory) where clause
• Replace db.close with ds.destroy

[KishanBagaria]

lol tried running this and then

fish: Job 1, '/Applications/Texts.app/Content…' terminated by signal SIGSEGV (Address boundary error)

[javiercr]

lol tried running this and then

fish: Job 1, '/Applications/Texts.app/Content…' terminated by signal SIGSEGV (Address boundary error)

fish: Job 1, '/Applications/Texts.app/Content…' terminated by signal SIGSEGV (Address boundary error)

it's actually a virus, got all your passwords now. jk, that's weird, let me test it with a production build, I just tested it in dev mode and opened the PR to document the thing.

[javiercr]

uhm, it's working fine for me (build branch > run production Texts.app > load the platform externally > restart).

[javiercr]

nvm, actually it crash when selecting a thread 🫥 (platform loaded externally)

[KishanBagaria]

its alright tho im using the dev environ w webpack to run this, no issues so far.

gonna run https://github.com/TextsHQ/platform-tests

[javiercr]

The same tests are failing in main than in this branch:

  Failed Tests: There were 4 failures

    sendMessage: image

      ✖ EntityNotFoundError: Could not find any entity of type "DBMessage" matching: { "id": "3EB02D261D573CF293EA|1", "threadID": "[email protected]" }


    deleteMessage

      ✖ Error: Message not found


    sendMessage: image

      ✖ EntityNotFoundError: Could not find any entity of type "DBMessage" matching: { "id": "3EB012CBE2602020EB3D|1", "threadID": "[email protected]" }


    deleteMessage

      ✖ Error: Message not found


  total:     30
  passing:   26
  failing:   4
  duration:  3.8s

Gonna dig into this errors, but this is unrelated to the PR

[KishanBagaria]

yep i described the issue here: https://linear.app/texts/issue/PLT-1033/whatsapp-sent-messages-arent-saved-in-db-before-return - a simple sleep fixes it

[javiercr]

In case anyone else want to test this, here is a build of this branch that can be easily installed:

CMD + J > Install platform integrations > Manifest URL:
https://drive.google.com/uc?id=1WZ0gyXRoTBpI6fpUV1KqR3CrEmWB2ui2

[javiercr]

This now includes #47 (which was pointing to this branch). @KishanBagaria let me know if you're happy to merge this as well.

[KishanBagaria]

yep hasn't broken for me yet