An enum is always the size needed to store its largest variant. Some of
the variants of the InformationElement and LteInformationElement are
substantially larger than the rest. Boxing the larger variants reduces
the size of the enum, in some cases by several kilobytes.
Since Rust does not currently support destructing a Box via pattern
matching, some code that destructures these enums had to be modified.
If rayhunter doesn't exit cleanly (e.g. during a battery outage), the
QMDL manifest may end up in a corrupted state. If that's the case,
rayhunter should try to recover by creating a new manifest. This'll let
it continue, and will preserve previous recordings, but they won't be
visible through the UI.
This commit refactors the config loading code to no longer require a
separate ConfigFile struct by taking advantage of serde's `default`
attribute. This causes serde to use the Config struct's default value
for that attribute for any missing attributes, which is what the
existing code was doing anyway.
This also fixes several clippy warnings.
Serde docs: https://serde.rs/container-attrs.html#default
* first pass at changing the UI color based on state
* adding flag to qmdl metadata for when hueristic is triggered
* update style for web page to match UI and have color alert on heuristic trigger
* add test analyzer
* rename example_analyzer to test_analyzer
* refactor ui update to not depend on server
* refactor to pass around color instead of display state for framebuffer channel
* add debug feature flag for test analyzer
* remove warning status from qmdl manifest
* dont keep has warning around
ND-JSON (newline-delimited JSON) is just a file with a list of JSON
objects separated by newlines. This way, as the analyzer harness
processes new packets, it can simply append JSON-serialized results
to a file without parsing the entire thing first.
Also simplifies the analysis stuff to all operate in the diag thread.
Mixing async and sync I/O leads to a multitude of complications, and
generally speaking it's much more convenient to stick to one paradigm
or the other. Since axum (and many other HTTP servers) use async,
and since async is a convenient model for performing operations like
"handle an MPSC message or file read, whichever happens first", let's
commit to an async interface.