You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
- When Pavex's Server fails to spawn a worker thread, capture that thread's name in the error message
14
+
- Add Server::try_serve to give users a chance to handle the case where no connection sources have been registered
15
+
- Cache accesses for path dependencies are now significantly faster, since 'cargo' is no longer in the hot path of determining the crate checksum
16
+
- Reduce the time spent generating JSON docs by leveraging 'cargo rustdoc' built-in caching mechanism for the single-crate usecase.
17
+
- Update Pavex's nightly to 2025-01-04. It unlocks the 2024 edition for Pavex users
18
+
- Enable backtraces for panics in the project template, being careful not to capture backtraces for normal errors in libraries such as anyhow. Update the command to force the toolchain install to work with the latest rustup version
19
+
20
+
### Fixed
21
+
22
+
- Use hyper_util's GracefulShutdown utility to ensure connections are indeed shut down gracefully
23
+
- Don't panic if we fail to serve a connection. Log the error as a warning.
24
+
- Lower the logging level to info when reporting a failure to establish a connection. This event kind is common enough not to deserve a higher logging level.
25
+
- Use the blueprint location as the project fingerprint rather than the workspace path, since a workspace may contain multiple blueprints, as it happens for Pavex's UI tests
26
+
27
+
### Other
28
+
29
+
- Migrate all tracing events that touch errors to `log_error!` to ensure consistent field naming
30
+
- Clarify why it's fine to panic if `tokio` can't create a current-thread runtime inside a Pavex server worker thread
31
+
- Improve suggested resolution for a missing nightly toolchain
32
+
- Re-export error-related fields in `pavex_tracing` from `tracing_log_error`
0 commit comments