Dereg individual

[forwardpointer]

Type

• Bug fix
• Feature addition
• Feature update
• Breaking change
• Non-functional change
• Documentation
• Infrastructure

Goals

Allow the deregistration of individual callbacks.

Technical Details

Since some callback function signatures are ambiguous, I decided to make separate types for each callback. These types are used internally by UwbConnector to keep track of what's what.

On the client side (i.e. UwbDevice and UwbSession), things largely stayed the same. They still fill out the structure of device callbacks or session callbacks, which they pass into Register[Device|Session]EventCallbacks. However, now those functions return a structure containing weak_ptr to the callback tokens corresponding to each of the callbacks they registered. If the client leaves some callbacks out of the structure, then the callback token for those callbacks will be nullptr (and the UwbConnector won't have added that token to its internal storage). The structure containing all session or device callbacks is nice to have in the interface because it allows the client to unambiguously label the callbacks as what they are, without needing to be aware of all the internal types UwbConnector uses to keep track.

Test Results

Regression tests pass

Reviewer Focus

Are there any other approaches to this that avoid the problem of having to define a type for each callback?

Future Work

• Actually call the deregistration in the TODO places
• Support recursive locks so that the callback can deregister itself.

Checklist

• Build target all compiles cleanly.
• clang-format and clang-tidy deltas produced no new output.
• Newly added functions include doxygen-style comment block.

[abeltrano]

Reviewer Focus

Are there any other approaches to this that avoid the problem of having to define a type for each callback?

An explicit type for each callback seems fine to me since there are actually different callback prototypes, so we really can't avoid them.

[abeltrano]

This is really well done. I added some comments with suggestions for consolidating stuff and simplifying the callback instances themselves, but otherwise this looks great.

[forwardpointer]

I just ran a regression test on the basic ranging scenario, and it passes. So it should be ready for merging now