Fix our Collection story
#196
Conversation
- Move from `MatchArgs` to `ObjectMatchRule` (note the "Object" prefix to avoid confusion with `zbus::MatchRule`) - Remove comment block on "Role fields" which are now a proper type `ObjectMatchRule` - Place the clippy suppression on the item that violates the rule. ToDo: - [ ] Document the methods properly This probably requires some digging in sources to find out what all arguments actually (were meant to) mean.
- `Live` is now known as `Politeness` - `ObjectMatchRule` not verifiable
`FromIterator` allows us to `.collect()` into an `InterfaceSet` `Default` impl returns an empty set.
- Remove `MatchArgs` - Move over `MatchType`, `TreeTraversalType` to `object_match.rs` - Introduce `ObjectMatchRule` with builder pattern - Move over / create tests
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #196 +/- ##
==========================================
+ Coverage 86.51% 87.22% +0.71%
==========================================
Files 39 40 +1
Lines 3344 3508 +164
==========================================
+ Hits 2893 3060 +167
+ Misses 451 448 -3 ☔ View full report in Codecov by Sentry. |
Missing item fixed.
Only `SortOrder::Canonical` and `SortOrder::ReverseCanonical` seem implemented. [upstream issue 140](Are all the AtspiCollectionSortOrder types really a thing?)
Seems to be unimplemented as our upstream libatspi peers say it is.
Move `SortOrder` to object_match module Delete duplicate tests.
|
Just took a cursory glance and am very excited! I can review this today, but probably still a couple hours out. Only thing I noticed is that there's link as to why no more then N objects will be returned from certain methods. Thank you Luuk! This looks like quite the upgrade! |
You raise a better point than I would like. In atk-adaptor of at-spi2-core, we find that #define MAX_CHILDREN 65536and both implemented sort-orders check for it. However, atk-adaptor may not be the only implementer on the bus. Accesskit does not implement
Thanks! What we need is a few experiments to see if things indeed work as we think it does. |
Remove maximum object number from documentation. Even though atk-adaptor implements the maximum as of present, they may not always have and it may not be the only implementation on the bus either.
atspi-common/src/object_match.rs
Outdated
|
|
||
| /// Insert a slice of `Interface`s | ||
| #[must_use] | ||
| pub fn interfaces(mut self, interfaces: &[Interface], mt: MatchType) -> Self { |
There was a problem hiding this comment.
Possibly consider interfaces: T where T: IntoIteer<Interface>? I think we've implemented that on InterfaceSet.
There was a problem hiding this comment.
Great idea - thanks!
| assert_eq!(rule.states, StateSet::new(State::Active)); | ||
| assert_eq!( | ||
| rule.attr, | ||
| [("name".to_string(), "value".to_string())].iter().cloned().collect() |
There was a problem hiding this comment.
Implemented!
Only `Canonical` and `ReverseCanonical` are in use in atk-adaptor. Other or future implementation might use other variants. (but this is unlikely at this point in time.)
Items may be either &Interface or Interface
`IntoIterator` uses `StateSet`s own iter, which returned `impl Iterator<Item = State>`, The `IntoIterator` implementation does not accept the opaque type, it needs the specific type to coincide with its own type, so the return type on `StateSet::iter` is adjusted to enumflags2::Iter<State>`.
atk-adaptor does not offer this method, but another maybe might. Added information in docs.
ToDo:
The above may take some digging in remote sources or asking around.
Note that 'collection.xml' isn't that helpful here.