Core event struct/data refactor #615
Conversation
* Separate header and data struct for the event. * Remove duplicate struct in split code.
petejohanson
changed the title
petejohanson
force-pushed
the
core/position-event-data-refactor
branch
4 times, most recently
from
4478f59 to
803361e
Compare
* Make it easier to use *just* event payloads by defining the data, and then having event manager macros generate "wrapper structs" * Improve is_*/cast_* APIs to hide details of full event struct. * Create `zmk_event_t` typedef to pass to event handlers. * Bring event names inline w/ consistent `zmk_` prefix.
petejohanson
force-pushed
the
core/position-event-data-refactor
branch
from
803361e to
420bf65
Compare
|
The last time |
This is one intended to be treated as opaque, you shouldn't ever be extracting fields from it directly. |
There was a problem hiding this comment.
Had a look through and didn't spot anything glaring.
The event names realignment could probably be a separate commit but it's controversial.
I'll defer to @joelspadin to comment on the change in approach as he'll have a better understanding.
E2E tests passed.
|
|
||
| #define ZMK_EVENT_RELEASE(ev) zmk_event_manager_release((struct zmk_event_header *)ev); | ||
| #define ZMK_EVENT_FREE(ev) k_free((void *)ev); |
There was a problem hiding this comment.
It is probably worth documenting somewhere that events must not contain any dynamically allocated memory, or it will leak.
petejohanson
force-pushed
the
core/position-event-data-refactor
branch
from
2f9b431 to
beae120
Compare
* Use a single `as_foo` generated function to conditionally return a certain event type from a generic `zmk_event_t*` pointer.
petejohanson
force-pushed
the
core/position-event-data-refactor
branch
from
beae120 to
2bc7df1
Compare
|
@innovaker Had some good feedback from @joelspadin, with just some minor tweaks since his last review. Ready for final review from you. |
This came up from freedback on #547, where @joelspadin had noted it would be nice to be able to copy data from event to type used locally.
A big part of the issue with doing that was the way event manager required that event types mix their data and the first "header" field.
This refactor removes that need, by having the event manager code generate a wrapper/containing type for each event type, leaving the data payload "pure data".
is_andcast_prefixes updated appropriately, and we've also made a typedefzmk_event_tto be passed to subscription callbacks.Once we're happy, I'll rebase #547