A few minor refactors, and a more major one that renames
RecordingStore's update_entry_qmdl_size to
update_current_entry_qmdl_size, since the only time we're ever updating
an entry's QMDL size is when it's the current one.
Major changes:
* QmdlWriter now outputs gzipped QMDL files by default
* QmdlReader renamed to QmdlMessageReader, and reads both compressed and
uncompressed QMDL. It no longer requires bounding to avoid reading
partially written files.
When there is a significant difference between the user's browser's time
and the system time, a button appears in the web UI to fix the system
time. This time will then be used to correct both data inside of PCAPs
and any metadata.
We don't actually set the system time to this value. Instead, rayhunter
adjusts any timestamps it handles by an offset. That offset defaults to
zero, and the user adjusts it by hitting the button in the web UI. The
main reason for this is device portability.
I haven't investigated whether it would actually be easy to set the real
system time. It's possible that it works the same way across all
devices.
These payloads would previous cause panic on underflow.
The fuzzing setup lives in
https://github.com/untitaker/rayhunter/tree/fuzz-wip -- I can eventually
upstream it though right now it runs very inefficiently.
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.
Will Greenberg