Don't process the same transaction twice

[MTRNord]

This implements the transaction code handling as described in https://spec.matrix.org/v1.2/application-service-api/#pushing-events

TODO

• Write Test(s)
• Check if this is actually smart or totally performance hell :)

Note

This does not fix the UB we have with the sync_token on the next step I think this was fixed meanwhile?

[MTRNord]

One thing to consider is that receive_transaction is moved over to the appservice crate. But not sure if that will actually work with the store as the appservice stuff is in another crate and pub(crate) would make it not visible to the appservice

[poljar]

One thing to consider is that receive_transaction is moved over to the appservice crate. But not sure if that will actually work with the store as the appservice stuff is in another crate and pub(crate) would make it not visible to the appservice

Is it? Where? In this PR?

AFAIR it can't live in the app-service crate unless we're willing to expose internal Client things, and this is a nice compromise which let's you receive responses without them being sent using the Client.

That being said, I'm not super familiar with app-services to say if that's the best approach.

[MTRNord]

One thing to consider is that receive_transaction is moved over to the appservice crate. But not sure if that will actually work with the store as the appservice stuff is in another crate and pub(crate) would make it not visible to the appservice

Is it? Where? In this PR?

AFAIR it can't live in the app-service crate unless we're willing to expose internal Client things, and this is a nice compromise which let's you receive responses without them being sent using the Client.

That being said, I'm not super familiar with app-services to say if that's the best approach.

Sorry I badly worded it. It was from the chat earlier and actually should say that it may be considered to do. Since then I came to the same conclusion though. Apart from the store thingy I added it cant be moved to the appservice crate.

[codecov]

Codecov Report

Merging #560 (e8318dc) into main (091fab8) will increase coverage by 0.26%.
The diff coverage is 97.56%.

@@            Coverage Diff             @@
##             main     #560      +/-   ##
==========================================
+ Coverage   77.76%   78.03%   +0.26%     
==========================================
  Files          92       92              
  Lines       13675    13885     +210     
==========================================
+ Hits        10635    10835     +200     
- Misses       3040     3050      +10     

Impacted Files	Coverage Δ
crates/matrix-sdk-appservice/tests/tests.rs	96.48% <96.66%> (+0.03%)	⬆️
crates/matrix-sdk/src/client/mod.rs	84.56% <100.00%> (-0.14%)	⬇️
crates/matrix-sdk-base/src/client.rs	78.45% <0.00%> (+1.13%)	⬆️
crates/matrix-sdk-base/src/rooms/normal.rs	81.35% <0.00%> (+1.59%)	⬆️
crates/matrix-sdk/src/room/common.rs	67.52% <0.00%> (+14.95%)	⬆️
crates/matrix-sdk-base/src/rooms/members.rs	50.00% <0.00%> (+17.85%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 091fab8...e8318dc. Read the comment docs.

[MTRNord]

Test fails seem unrelated?

Binding test was a stack overflow on setup and sso-login tests were a Connection reset error while running them

[jplatte]

Re-running failed jobs…

[johannescpk]

LGTM, just a small nit.