Yehowshua/ZIm
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
22249 commits 1103 branches 1111 tags 398 MiB
Commit graph

519 commits

Author SHA1 Message Date
Petros Amoiridis
73af155dd6
Refactor Database::remove_contact 
Refactor it to avoid sending irrelevant messages to update the UI.

Co-Authored-By: Antonio Scandurra <me@as-cii.com>
 2023-01-26 19:01:51 +02:00
Petros Amoiridis
5d4eb2b7ae
Push responder and requester to remove_contacts 
When we ask the server to remove a contact we need to push the requester and responder ids to `remove_contacts` so that when the UI updates, the correct contacts will disappear from the list.

Co-Authored-By: Antonio Scandurra <me@as-cii.com>
 2023-01-25 13:10:29 +02:00
Antonio Scandurra
8487ae77e7 Share new worktrees when resharing project 
Co-Authored-By: Nathan Sobo <nathan@zed.dev>
 2023-01-06 12:58:19 -07:00
Nathan Sobo
789bbf15b7 Update buffer files when synchronizing buffers 
It's possible that the host was disconnected when attempting to notify
guests of a file save, so we need to transmit this in order to correctly
update the file's mtime.

Next failing seed OPERATIONS=200 SEED=6894
 2023-01-04 12:33:48 -07:00
Nathan Sobo
90fb9b53ad WIP 2023-01-03 13:30:14 -07:00
Max Brunsfeld
599acf0daa WIP - Panic immediately when detecting non-determinism via a change to the execution trace 2022-12-23 17:34:13 -08:00
Antonio Scandurra
47348542ef Synchronize buffers when either the host or a guest reconnects 2022-12-21 14:20:56 +01:00
Max Brunsfeld
213be3d6bd Delete stale projects after cleanup interval, via server foreign key cascade 2022-12-20 17:27:42 -08:00
Max Brunsfeld
cb8962691a Remove unnecessary UnshareProject message sent to clients leaving a project 2022-12-20 16:58:44 -08:00
Antonio Scandurra
52babc51a0 Make host reconnection test pass when mutating worktree while offline 2022-12-20 17:30:58 +01:00
Antonio Scandurra
1a3940a12e Fix project reconnection test to ensure rooms actually reconnects 2022-12-20 14:51:46 +01:00
Antonio Scandurra
1aec691b35 Sketch out project reconnection routine on the server 2022-12-20 12:03:43 +01:00
Max Brunsfeld
ad37034960 Identify LiveKit room participants by user id, not peer id 
This way, their participant id can remain the same when they reconnect.
 2022-12-15 17:19:32 -08:00
Antonio Scandurra
5a334622ea 💄 2022-12-15 16:34:59 +01:00
Antonio Scandurra
86e5ae1f2e Allow nulls in projects.host_connection_{id,server_id} 
The server version on stable won't be able to fill values for those
columns when we deploy the migration to preview.

With this commit we're also dropping the unused `worktree_extensions`
and `project_activity_periods` tables. The last version of the server
on stable (0.2.6) doesn't contain any code that accesses those tables.
 2022-12-15 11:30:51 +01:00
Antonio Scandurra
067a19c971 Avoid logging an error when user who hasn't joined any room disconnects 2022-12-15 10:45:03 +01:00
Antonio Scandurra
688f179256 Use "id" nomenclature more consistently 2022-12-15 10:15:59 +01:00
Max Brunsfeld
6c58a4f885 Fix stale server queries, use foreign keys from connectionsn to servers 2022-12-14 17:34:24 -08:00
Antonio Scandurra
363e3cae4b WIP 2022-12-14 19:25:07 +01:00
Antonio Scandurra
930be6706f WIP 2022-12-14 18:02:39 +01:00
Antonio Scandurra
05e99eb67e Introduce an epoch to ConnectionId and PeerId 2022-12-14 15:55:56 +01:00
Antonio Scandurra
674fddac87 Instrument rpc::Server::start and reduce cleanup timeout again 2022-12-14 11:42:12 +01:00
Antonio Scandurra
9530976f61 Try using a longer timeout for cleaning up stale rooms 2022-12-14 11:24:36 +01:00
Antonio Scandurra
897506c797 Define readiness probe to know when the new server can accept traffic 2022-12-14 08:54:46 +01:00
Antonio Scandurra
dde6cf596e Don't wait for stale project deletion before listening for connections 2022-12-14 08:42:34 +01:00
Max Brunsfeld
45bfcfc3b8 Fix excessive delay before clearing stale room data 
Co-authored-by: Antonio Scandurra <antonio@zed.dev>
 2022-12-13 11:17:20 -08:00
Max Brunsfeld
6ad326ac58 Wait longer before deleting outdated rooms from the database 
Co-authored-by: Antonio Scandurra <antonio@zed.dev>
 2022-12-13 10:14:43 -08:00
Antonio Scandurra
4dc838fbb7 Reset connection pool when tearing down the server in tests 2022-12-13 13:51:25 +01:00
Antonio Scandurra
d4c8fa3090 Use a synchronous mutex for ConnectionPool 2022-12-13 13:50:51 +01:00
Antonio Scandurra
0220d7ba5d Include room_id in CallCanceled message 
This ensures we don't accidentally cancel old calls.
 2022-12-13 11:43:09 +01:00
Antonio Scandurra
e2b132ef23 💄 2022-12-13 11:37:39 +01:00
Antonio Scandurra
7e8d9d52d3 Delete stray debug statement 2022-12-13 11:36:40 +01:00
Antonio Scandurra
6a6a032f1f Delete stale rooms/participants after RECONNECT_TIMEOUT 2022-12-13 11:32:37 +01:00
Antonio Scandurra
beea9b68ff Allow re-joining room after server restarts 2022-12-12 16:03:21 +01:00
Antonio Scandurra
456396ca6e Rename connection_lost to answering_connection_lost 2022-12-12 11:43:08 +01:00
Antonio Scandurra
895c365485 Introduce random reconnections in the randomized test 2022-12-12 11:43:08 +01:00
Antonio Scandurra
aca3f02590 Re-join room when client temporarily loses connection 2022-12-12 11:43:08 +01:00
Antonio Scandurra
d74fb97158 Remove Executor trait from collab and use an enum instead 
This will let us save off the executor and avoid using generics.
 2022-12-12 11:43:08 +01:00
Antonio Scandurra
fc7b01b74e Fix busy status when accepting a contact request 
Previously, we would send an contact update when accepting a request
using the same `busy` status for both the requester and the responder.
This was obviously wrong and caused the requester to see their own
busy status as the newly-added responder contact's status.
 2022-12-06 10:19:34 +01:00
Antonio Scandurra
7bbd97cfb9 Send diagnostic summaries synchronously 2022-12-05 19:07:06 +01:00
Antonio Scandurra
5443d9cffe Return project collaborators and connection IDs in a RoomGuard 2022-12-05 18:37:01 +01:00
Antonio Scandurra
d96f524fb6 WIP: Manually rollback transactions to avoid spurious savepoint failure 
TODO:
- Avoid unwrapping transaction after f(tx)
- Remove duplication between `transaction` and `room_transaction`
- Introduce random delay before and after committing a transaction
- Run lots of randomized tests
- Investigate diverging diagnostic summaries

Co-Authored-By: Max Brunsfeld <max@zed.dev>
 2022-12-02 20:36:50 +01:00
Antonio Scandurra
48b6ee313f Use i32 to represent Postgres INTEGER types in Rust 2022-12-02 13:58:54 +01:00
Antonio Scandurra
dec5f37e4e Finish porting remaining db methods to sea-orm 2022-12-02 13:58:23 +01:00
Antonio Scandurra
29a4baf346 Replace i32 with u32 for database columns 
We never expect to return signed integers and so we shouldn't use
a signed type. I think this was a limitation of sqlx.
 2022-12-01 17:47:51 +01:00
Antonio Scandurra
d2385bd6a0 Start using the new sea-orm backed database 2022-12-01 14:41:59 +01:00
Antonio Scandurra
cd0b663f62 Introduce per-room lock acquired before committing a transaction 2022-11-28 17:00:47 +01:00
Antonio Scandurra
c3d556d9bd Don't take an Arc<Server> in message handlers 2022-11-18 11:45:42 +01:00
Antonio Scandurra
44bb2ce024 Rename Store to ConnectionPool 2022-11-17 19:03:59 +01:00
Antonio Scandurra
6c83be3f89 Remove obsolete code from Store 2022-11-17 18:46:39 +01:00