Add CryptoApi.encryptToDeviceMessages() and deprecate Crypto.encryptAndSendToDevices() #4380
Conversation
Deprecate Crypto. encryptAndSendToDevices and MatrixClient. encryptAndSendToDevices
hughns
added
T-Enhancement
T-Deprecation
hughns
changed the title
hughns
changed the base branch from
develop
to
hughns/matrix-sdk-crypto-wasm-8
hughns
changed the base branch from
hughns/matrix-sdk-crypto-wasm-8
to
develop
src/crypto-api/index.ts
Outdated
| * set of devices. | ||
| * | ||
| * @param eventType the type of the event to send | ||
| * @param devices an array of (user ID, device ID) pairs to encrypt the payload for |
There was a problem hiding this comment.
I wonder if we can/should impart any advice about the maximum number of entries in devices?
There was a problem hiding this comment.
A cursory look suggests that the limiting factor would be the size of the claim request generated by KeyClaimManager.ensureSessionsForUsers().
Might it make sense to batch that at some sensible level? Similar to the batch size of 20 that is imposed for actually sending the to-device events to the server.
As a user of the higher-level CryptoApi I think any batching would be done for me automatically (which currently isn't the case).
There was a problem hiding this comment.
It would be nice to add an integration test in spec/integ/crypto that exercises the encryption without mocking out bits of OlmMachine. I'm aware that such things can be fiddly to get to work, though, so I won't insist.
|
BTW it looks like the description in the PR is a bit outdated -- could you give it a refresh? |
hughns
changed the title
Fixes #3304
Depends on matrix-org/matrix-rust-sdk-crypto-wasm#140 and #4396
Checklist
public/exportedsymbols have accurate TSDoc documentation.