Getting full panics here in prod:
thread '<unnamed>' (23) panicked at /usr/local/cargo/registry/src/index.crates.io-1949cf8c6b5b557f/dioxus-fullstack-0.7.0/src/magic.rs:554:22:
called `Result::unwrap()` on an `Err` value: Response { status: 500, version: HTTP/1.1, headers: {"content-type": "application/json"}, body: Body(UnsyncBoxBody) }
Could be triggered by something as simple as a client choosing to send formdata-style body instead of JSON. (I discovered this because exactly that was happening; not maliciously, but because of the upgrade from 0.6.3 to 0.7).
As far as I can tell, this was just meant to be an early return with a failure case of the Response, mapped from the ServerFnError. The .unwrap() left there was probably just an erroneous artifact from the various waves of the development process.
If this is incorrect or I've missed something, please let me know! But I believe this is a very simple fix for an otherwise critical problem
This flag allows users to get the raw JSON output from cargo instead
of having it processed by Dioxus. This is useful for integration with
other tools that expect cargo's native JSON diagnostic format.
When enabled, cargo's JSON messages are relayed directly to stdout
while still allowing DX to track essential build information internally.
Usage: dx build --raw-json-diagnostics
* fix(fullstack-macro): allow server functions to be called without HTTP context
Server functions can now be invoked directly from server-side code without
requiring an HTTP request context. When called outside an HTTP handler,
extractors will use default values instead of failing, matching the behavior
of the old server_fn implementation.
Closes#4907
* fix(fullstack-macro): allow server functions without extractors to be called from backend
Functions with no server extractors use Default::default() and work from backend code.
Functions with extractors still require HTTP context.
Closes#4907
* Fix dioxus-native to run without the net feature
* Fix local assets to be available without net features
* fix: compile out the inner_net_provider field entirely
* feat: make dioxus-native run without html feature
* Enhance Android log handling in AppBuilder
- Introduced adb_logcat fields to manage Android logcat output.
- Implemented logcat spawning and parsing to differentiate log levels.
- Updated push_stdio method to filter Android logs based on trace settings.
- Added utility functions for retrieving app PID and parsing logcat levels.
* automatic retry to get pid
* pipe RustStdoutStderr as well
* strip logcat prefix
* filter in rust instead
* Refine log filtering in Rust output handling
- Improved log level detection by checking for prefixes after the colon in messages.
- Enhanced filtering to only show logs with standard tracing levels or fatal errors.
- Streamlined the process of stripping logcat metadata while preserving message content.
* cleanup output and kill proc
* capture logs from ios device
* clippy
---------
Co-authored-by: Jonathan Kelley <jkelleyrtp@gmail.com>
* Add more read/write methods for string signals
* add helpers for HashMap and HashSet as well
* fix typo
* fix conflicting string and str extensions
* fix counters example
* fix(hotpatch): Gracefully handle missing '-flavor' flag during WASM link
* drops logs
---------
Co-authored-by: Jonathan Kelley <jkelleyrtp@gmail.com>
* add component subcommand
* add dx component add, remove and list commands
* handle rust dependencies
* handle component dependencies
* fetch remote registries
* sort components before listing them
* remove local dependency
* move te component manifest into the dioxus workspace
* add additional instalation instructions the first time the component
module is created
* fix mod name
* fix clippy
* improve docs and move to the main branch of components
* fix typos
* more typos
* undo const-serialize changes
* remove a few unwraps
* Allow adding multiple components at once and add a `--all` flag
* add options to the config for the base path, global asset path and
default registry
* component_path -> component_dir
* component_dir is always relative to the crate root
* fix clippy
* fix hot patching with a base path
* read the hot patch file from stdin instead of a CLI arg to allow for longer inputs
* fix request imports
* wip: code style changes
* rename to "components" instead of "component"
* bump lock
* move to method on request
* read until eof in the case of pretty json
* strip is now copy
---------
Co-authored-by: Jonathan Kelley <jkelleyrtp@gmail.com>
* add component subcommand
* add dx component add, remove and list commands
* handle rust dependencies
* handle component dependencies
* fetch remote registries
* sort components before listing them
* remove local dependency
* move te component manifest into the dioxus workspace
* add additional instalation instructions the first time the component
module is created
* fix mod name
* fix clippy
* improve docs and move to the main branch of components
* fix typos
* more typos
* undo const-serialize changes
* remove a few unwraps
* Allow adding multiple components at once and add a `--all` flag
* add options to the config for the base path, global asset path and
default registry
* component_path -> component_dir
* component_dir is always relative to the crate root
* fix clippy
* fix request imports
* wip: code style changes
* rename to "components" instead of "component"
* bump lock
---------
Co-authored-by: Jonathan Kelley <jkelleyrtp@gmail.com>
* Add optional static_dir support
- add optional application.static_dir field defaulting to none
- copy configured static directory into web output during build/serve
- document config knob and add unit tests
- force full rebuild when static_dir contents change
- sync removes stale outputs and logs copy activity
- track static manifest next to platform dir
Tests: cargo fmt
* default to public dir to match vite
* static_dir -> public_dir
* wip
* deser as none...
* wip... use .manifest.json
* implement dumb form of asset manifest generation
* implement manifest using asset system directly
* cleanup
* dont create extra asset dir for server
---------
Co-authored-by: Evan Almloff <evanalmloff@gmail.com>
Co-authored-by: Jonathan Kelley <jkelleyrtp@gmail.com>
Scenario: without this patch, if you fail a build, fix the code, let the thin build succeed, and then refresh the browser... then after reloading the browser page, it gets stuck on the "One sec" loader because the devserver status never clears the earlier BuildError.
- mark build status ready after sending hot patch messages so refreshes succeed
- reuse the same helper from reload command to avoid stale loading screens
- add set_ready() helper to avoid duplicate status broadcasts
* wip: redirect
* redirect example
* fix file upload not in a form
* wip
* deny missing cfgs
* wip: add router fallback but it's not working
* fix suspense bug after rebuilds that queued work
* more cleanup
* add test for error codes
* add helpful comment
* bring back middleware annotation
* fix cfg
* add server feature
* fix doctest
* just disable liveview test for now
* fix liveview
* dont send init
* cache more playwright!
* add more pkgs to cache
* allow `--server` only builds!
* rollback example
* capture `/:route` routes
* fix: properly implement file uploads in forms
* implement desktop form uploads
* Performant serialization of form bytes
* finish up file stuff
* cleanups
* add request headers to streaming context
* remove some unwraps
* fix: hotpatching fullstack
* add hot-patching fullstack test
* add caching to playwright apt-get
* only run playwright on beta
* install binutils?
* dont commit lockfile
* ignore lockfiles
* make sure ld is in path
* Look for strip settings in inherited profiles and global configs
* fix clippy
* use cargo home env var if it exists
* fix cargo home config path
* small stylistic changes
---------
Co-authored-by: Jonathan Kelley <jkelleyrtp@gmail.com>
* Fix route hydration with ssg without early chunk commit
* fix clobbered port for fullstack spread (merge conflict)
* wait for hydration
---------
Co-authored-by: Jonathan Kelley <jkelleyrtp@gmail.com>
Craig