The previous code had a number of problems, including:
- Calls to the filters API were scattered through UI and viewmodel code.
- Repeated places where the differences between the v1 and v2 Mastodon
filters API had to be handled.
- UI and viewmodel code using the network filter classes, which tied
them to the API implementation.
- Error handling was inconsistent.
Fix this.
## FiltersRepository
- All filter management now goes through `FiltersRepository`.
- `FiltersRepository` exposes the current set of filters as a
`StateFlow`, and automatically updates it when the current server
changes or any changes to filters are made. This makes
`FilterChangeEvent` obsolete.
- Other operations on filters are exposed through `FiltersRepository` as
functions for viewmodels to call.
- Within the bulk of the app a new `Filter` class is used to represent a
filter; handling the differences between the v1 and v2 APIs is
encapsulated in `FiltersRepository`.
- Represent errors when handling filters as subclasses of `PachliError`,
and use `Result<V, E>` throughout, including using `ApiResult` for all
filter API results.
- Provide different types to distinguish between new-and-unsaved
filters, new-and-unsaved keywords, and in-progress edits to filters.
## Editing filters
- Accept an optional complete filter, or filter ID, as parameters in the
intent that launches `EditFilterActivity`. Pass those to the viewmodel
using assisted injection so the viewmodel has the info immediately.
- In the viewmodel use a new `FilterViewData` type to model the data
used to display and edit the filter.
- Start using the UiSuccess/UiError model. Refrain from cutting over to
full the action implementation as that would be a much larger change.
- Use `FiltersRepository` instead of making any API calls directly.
## Listing filters
- Use `FiltersRepository` instead of making any API calls directly.
## EventHub
- Remove `FilterChangedEvent`. Update everywhere that used it to use the
flow from `FiltersRepository`.
Abstract common CI setup tasks (setting up Java, Gradle, etc) in to a
single action that can be used by all CI workflows.
Run the lint, test, and assemble CI tasks in parallel for each variant
rather than in series, which cuts ~ 7 minutes (approx. 50%) off the CI
runtime.
Update code in checks and core/navigation to fix new tests.
New lint rules highlighted a potential crash; the use of named match
groups (used here when extracting server versions) requires API >= 26 or
throws an exception.
Use the group numbers instead of names when extracting the value, but
keep the group names in the regular expressions for readability.
Mastodon counts post lengths by considering emojis to be single
characters, no matter how many unicode code points they are composed of.
So "😜" has length 1.
Pachli was using `String.length`, which considers "😜" as length 2.
Correct the calculation by using a BreakIterator to count the characters
in the string, which treats multi-character emojis as a length 1.
Poll options had a similar problem, exacerbated by the Mastodon web UI
also having the same problem, see
https://github.com/mastodon/mastodon/issues/28336.
Fix that by creating `MastodonLengthFilter`, an `InputFilter` that does
the right thing for regular text that may contain emojis.
See also https://github.com/tuskyapp/Tusky/pull/4152, which has the fix
for status length but not polls.
---------
Co-authored-by: Konrad Pozniak <opensource@connyduck.at>
The existing code base is a single monolithic module. This is relatively
simple to configure, but many of the tasks to compile the module and
produce the final app have to run in series.
This is unnecessarily slow.
This change starts to split the code in to multiple modules, which are:
- :core:account - AccountManager, to break a dependency cycle
- :core:common - low level types or utilities used in many other modules
- :core:database - database types, DAOs, and DI infrastructure
- :core:network - network types, API definitions, and DI infrastructure
- :core:preferences - shared preferences definitions and DI
infrastructure
- :core:testing - fakes and rules used across different modules
Benchmarking with gradle-profiler shows a ~ 17% reduction in incremental
build times after an ABI change. That will improve further as more code
is moved to modules.
The rough mechanics of the changes are:
- Create the modules, and move existing files in to them. This causes a
lot of churn in import arguments.
- Convert build.gradle files to build.gradle.kts
- Separate out the data required to display a tab (`TabViewData`) from
the data required to configure a tab (`TabData`) to avoid circular
dependencies.
- Abstract the repeated build logic shared between the modules in to
a set of plugins under `build-logic/`, to simplify configuration of
the application and library builds.
- Be explicit that some nullable types are non-null at time of use.
Nullable properties in types imported from modules generally can't be
smart cast to non-null. There's a detailed discussion of why this
restriction exists at
https://discuss.kotlinlang.org/t/what-is-the-reason-behind-smart-cast-being-impossible-to-perform-when-referenced-class-is-in-another-module/2201.
The changes highlight design problems with the current code, including:
- The main application code is too tightly coupled to the network types
- Too many values are declared unnecessarily nullable
- Dependency cycles between code that make modularisation difficult
Future changes will add more modules.
See #291.
Nik Clayton
renovate[bot]
