Releases: get-convex/convex-backend
Releases · get-convex/convex-backend
Precompiled 2024-04-18-4c52b74
Wire up queries to isolate2 (#24784) GitOrigin-RevId: 10d1b67cdd8ff9ba38822d8563ae17beb0af1382
Precompiled 2024-04-17-f9a68b5
Wire up error handling to isolate2 (#24731) GitOrigin-RevId: 7214af8b4ad6faf6c204feec96b16d1009052763
Precompiled 2024-04-16-0d6c4ca
Remove a bunch of unecessary fallible conversions (#24678) GitOrigin-RevId: 25c1bf89e85737c0f0c36af88bc0f54d6d30dd9f
Precompiled 2024-04-15-addaff2
precompiled-2024-04-15-addaff2 Add OpProvider for sharing op implementations between isolate1 and is…
Precompiled 2024-04-12-ef15827
[Tracing] Add tracing to functions in `Isolate::run` (#24558) There are some missing spans in `Isolate::run` so this aims to capture some of the missing execution time. GitOrigin-RevId: baea64010a124a53fac037cfaa4e3069490f2189
Precompiled 2024-04-11-4496d46
[Fix] Shutdown + join futures on `retention_manager` shutdown (#24515) It turns out that if we don't join a handle after shutdown, it can still perform work after the worker that created it has been dropped. So, I refactored the retention manager code to join all of its handles on shutdown. Its slightly concerning that we use this same pattern of calling `.shutdown()` in other places without joining the future because we can't guarantee that the handle stops doing work before the creator goes away. Should we try to get rid of this pattern entirely? Intuitively speaking it would seem that calling `.shutdown()` would terminate a thread synchronously, but this is not actually what happens. GitOrigin-RevId: db38a2e748ebd958f6be2cde8258f7ae2f1ce00e
Precompiled 2024-04-10-bb2197b
Make the dashboard handle structured log lines (#24131) The dashboard version of https://github.com/get-convex/convex/pull/24129. Note that we still allow unstructured log lines too. There should be no user visible changes here. GitOrigin-RevId: 731248c2c930d35b641f209b07dd21d808513578
Precompiled 2024-04-09-e0f2ff2
Add request tracing middleware. (#24427) * Add request_id extractor. * Set the request_id to sentry scope via middleware. * Enable tracing based on middleware. GitOrigin-RevId: 8d891352864a3fc599a3ebdcd59fad0d83b15d83
Precompiled 2024-04-08-aa99e12
[cluster migration] fix parallel futures (#24411) GitOrigin-RevId: 1d571d3294aa7d5343d2fa79ca18784dbfca199c
Precompiled 2024-04-05-1f403e7
Fixed incorrectly bumping retention confirmed deleted ts (#24332) Made it such that we don't bump the confirmed deleted timestamp of document retention on dry runs GitOrigin-RevId: e5576a096c65195f6437fa7072896616695f4d7f