Viewport was not scrolled back to the place were change was made when
undoing global operation (refactoring).
GitOrigin-RevId: f44c6279fbae1ca72373ae3574234e0d609df0ef
- for a state-query that never has more than one valid match at the same time ( query { } ) it is equivalent to Query.collect
- for a query that can potentially have many valid matches (Entity.each()) it is almost definitely an error because only one of them will be processed
GitOrigin-RevId: b9fce3aa7773f8c1642439260cf0dbd1448efcad
it used to update the threadLocal in place, but this is fragile because next updateThreadContext will reset it to some past value
GitOrigin-RevId: 2fd4ba8aa8b6b481a1648cce5f919434c119aebb
because it does not wait for dbsource to catch up, it will show the consumer a glimpse of the future that will be gone when DbContextElement updates the thread context
anyway it is inconsistent because it only moves forward and provides no consistency guarantees, the only way to safely consume match's value is to observe it inside withMatch
GitOrigin-RevId: 33d7096fff8fc118cf092efbcb24c9879bcc0660
it was testing a legacy api that was migrated to rete a long time ago, rete has its own tests
nitpick: wait for rete to catch up to db source, not transactor
GitOrigin-RevId: b6c078b529e1766435e62f8124e1db874e3a9842
[fleet] set desktop window manager to 0.0.8
[fleet] files generated by gradle
[fleet] fix typos
[fleet] fix jcef typing with kdt
[fleet] add clicks count field; store window manager evetn instead of awt event; now mouse works in JCEF event with KDT
[fleet] reduce the number of buffer copying
[fleet] add an utility function to access direct byte buffer pointer
it's used by JCEF
[fleet] invalidate screen property
[fleet] rename
[fleet] fix stucked modifiers
[fleet] add configuration to run with desktop window manager
[fleet] use atomic to be safe in case if `requestFrame` is called from non ui threads
[fleet] fix ctrl-c in terminal with desktop window manager
[fleet] drop the frame-requested flag immediately after the check
otherwise it might overwrite next request if it was scheduled during frame rendering
[fleet] synchronize reads with writes in ime in a propper way
[fleet] sync gradle with iml
[fleet] fix after rebase
[fleet] use non mutable editor for reading
[fleet] fix arrow in the terminal
[fleet] add a basic implementation of the text client interface for the terminal
[fleet] cleanup
[fleet] add a comment
[fleet] input methods expect that any mutable editor transaction will be executed at some point
hopefully we will remove this requirement later
[fleet] improve IME
[fleet] fix togglable application menu items; fix fullscreen toggle
[fleet] fix modifiers key press
[fleet] we don't use NSEvent timestamp at the moment
[fleet] minor
[fleet] synchronously read editor state
[fleet] make writes async
[fleet] wip
[fleet] according to docs text range might be out of document boundaries
[fleet] fix ime popup positioning
[fleet] more methods in the text client interface
[fleet] wip2
[fleet] wip3
[fleet] it works
[fleet] enable frost glass effect
[fleet] change jframe background more forcibly
though still not always working for windows other than first
[fleet] removed unused attribute
[fleet] removed because we already set window background in `BasicWindow`
[fleet] clipboard is implemented desktop toolkit
[fleet] sync theme with OS
[fleet] add window background effect property to window manager
[fleet] disable transition animation for window; otherwise it moves too slowly when dragged out
[fleet] window drag improvements
fix typo, now we send it cursorLeftWindow
don't use real event timestamps, to be closer to AWTWindowManager.
before the change mouse up was ignored in WindowDragAndDrop (see `isRecent`)
[fleet] cursor icons
# Conflicts:
# fleet/noria/windowManagement/implNative/srcJvmMain/noria/windowManagement/nativeImpl/DesktopMouseCursors.kt
[fleet] fix rendering stuck when some interactivity goes in terminal
[fleet] add implNative as dependency to DesktopDock
[fleet] add implNative module back
[fleet] move native wm sources
[fleet] set KDT to snapshot (do not merge in master!)
GitOrigin-RevId: a1eb27a63ca1056ddec562be1cb42aeaba913c5f
Enhancements:
- Add support for PTY for test execution
- Display & rerun tests from multiple tasks in single test tab
- Allow debugging tests from test runner
- Make rerun items in test runner menus reactive (prefetch tasks, enable/disable menu items as available tasks change)
- Make test navigation more robust if test cannot be located (fall back to ancestors)
- Improve stability of rerunning (reuse implicit task modifiers, use largest available subset of modifiers, etc.)
- Support optional groupings of tests (togglable via menu, e.g., for Gradle Test executors)
Other changes:
- Improve responsive landscape/portrait layout of test view
- Introduce new `testTaskProvider` API for rerunning tests
- Miscellaneous UI refactorings
- Determine state of root notes in test tree based on final state (exit code) of associated launch step entity
Known regressions (will be implemented later):
- Rerun tests in same tree currently does not work (every rerun spawns a new test tree)
- Selecting tests in node will not reduce output pane to output of selected test(s)
GitOrigin-RevId: 537f9f755f7303ea53f6a22baf744c8d096d8463
apparently it is possible that something will set thread bound database to a later version, which will short circuit waitForDbSourceToCatchUpWithTimestamp
thread bound database will be later updated by the first change in the coroutineContext because DbSource.ContextElement will be invoked and it will bind the database to the latest version it knows
fixes FL-33239 HB issue: DB timestamp becomes lower after leaving the spannedScope block
GitOrigin-RevId: 488ddedcb2ec5c7661a9cbf6d4f094425cc21030
it is possible for frontend to have an entity type that is not accessible from the workspace layer, but is stored in the shared partition with the workspace storage key
the idea is that workspace will populate the shared partition with the snapshot contents using raw json objects for values and then frontend will reify these entities and deserialize all the attributes. deserialization happens either when the transaction is applied if we already have a serializer, or on plugin loading
when running in short circuit mode, hackyNonBlockingChange might
capture a version of the database that does not contain an entity type
with its serializers, but will be applied to a version that already
has it. the change will be expanded into a plain assertion of json
values that will never be deserialized later
FL-33229 "Open" doesn't work from the welcome screen
GitOrigin-RevId: d5f87eab099d2be273f203f8cf43d3a9ab8c6018
