Conversation
Add CI workflow for Linux with build and test steps.
Removed xvfb installation and usage for running tests.
windowapi support for x11 and wayland
Pressing Ctrl+A followed by Backspace started a 300 ms fileSaveTimer instead of pushing the deletion onto the undo stack immediately. If the user pressed Ctrl+Z within that window the command had not been queued yet, so undo either did nothing or reverted an earlier operation. MkEdit: - Flush any pending edit at the very top of keyPressEvent, before undoData.editType is overwritten, so the real edit type is still available to updateRawDocument / postUndoSetup. - Treat a selection covering the whole document as a multi-block edit even when it sits on a single block. Previously Ctrl+A on a one-line document was classified as singleEdit, which stored only the current block in the undo snapshot and could not restore the deleted content. - Force a fresh preUndoSetup() snapshot for multiDelete regardless of whether fileSaveTimer is already running, so the pre-delete raw text is always captured. - Keep multiDelete out of removeAllMkData in clearMkEffects so the selection survives until QTextEdit::keyPressEvent runs and the whole selection is deleted in one step. - Add clampDocPos() and apply it wherever a cursor position is restored from selectRange (applyMkEffects, postCursorPosChangedSignal, showSelectionAfterRedo, setPostArrowKeys, restoreTextCursor, clearMkEffects, insertFromMimeData) to stop the "QTextCursor::setPosition: Position ... out of range" warnings after the document shrinks. - Sync MkEdit's selectRange to undoRedoSelectRange before applyMkEffects reads it on both Ctrl+Z and Ctrl+Y. MkTextDocument: - Add EditType::multiDelete to the enum and handle it in EditCommand's constructor, undo() and redo() alongside multiEdit / checkbox / enterPressed, so the command actually restores the full raw document.
reconstruct the func for style setting.
It is hard too test code with catchTest.
style change cannot use
TODO: global style reload in settingsdialog.cpp have a nice day~
CatchTests/CMakeLists.txt;Enabled CatchTests in CMakeListsl.txt
Aimed to maintain the project with Linux adoption alone
Repair the CatchTests target so it actually compiles, link the missing translation units, and complete/fix the four test files against the current behaviour of the classes under test. CMakeLists.txt - Add CMAKE_AUTOUIC_SEARCH_PATHS so .ui files in the project root are found from the tests subdirectory; ui_mainwindow.h and ui_settingsdialog.h were never generated before. - Add togglebutton.cpp and textsearchworker.cpp: views_handler.cpp instantiates ToggleButton and owns a TextSearchWorker, so the test binary failed at link time without these. - List mainwindow.ui and settingsdialog.ui explicitly for AUTOUIC. - Leave X11 linking commented out with a note, since windowapi.cpp is not part of the test target. MkEdit_Tests.cpp - Uncomment #include <catch2/catch.hpp>; the file was previously un-compilable. - Rewrite "bold double asterisk" and "bold double underscore" to route text through MkTextDocument with markdown enabled. The originals used a bare MkEdit backed by a plain QTextDocument and compared the input string against itself, so they passed without exercising any markdown logic. - Keep every other test body unchanged to preserve the existing pass/fail baseline. mktextdocument_tests.cpp - Add explicit includes for QChar, QString, QTextCursor and QTextCharFormat instead of relying on transitive includes. - Document the two behaviours the tests rely on: setMarkdownHandle(true) hides every block, and cursorPosChangedHandle(&range) unhides only the block at range.currentBlockNo (out-of-range indices are silent no-ops). Comment the deliberately out-of-range currentBlockNo assignments so they are not "fixed" later. - No test logic or expectation changes. NavigationView_Tests.cpp - Fill in the original empty "NavigationView Add file Test" placeholder. - Add coverage for addFolder, setVaultHandler, openFileFolder, copyFileFolderPath, renameFile (no index), deleteFile (no selection), keyPressEvent (Backspace and Return), expandTimerHandler, rowClicked, setRowsEditable and closeEditor. - Add a fixture that wires NavigationView to a real QFileSystemModel and NavigationProxyModel rooted at a QTemporaryDir for tests that need a valid index. - Observe signals via QObject::connect + lambdas rather than QSignalSpy, because the project's signals use non-const reference parameters. - Deliberately exclude filesystem side effects and inline-editor lifetime; those belong to NavigationProxyModel and to a UI-driven test, respectively. RecentFilesDialog_Tests.cpp - Replace the non-existent #include <QtTest/QTestKeyEvent> with <QtTest/QtTest>; the file did not compile before. - Add missing includes for QListWidget, QListWidgetItem and QScopedPointer. - Remove every delete listPtr: the dialog re-parents the supplied QListWidget in its constructor and owns it, so deleting it from the test was a use-after-free. - Fix "adding 2 paths ... with show()" to actually insert item2; the original declared it but never used it, contradicting the test name. - Send Tab with Qt::ControlModifier in all cases. A bare Tab can be consumed by QWidget::event()'s focus traversal before keyPressEvent() runs, and whether it is consumed depends on the dialog's focus state, which is platform-dependent because of the Qt::ToolTip window flag. RecentFilesDialog::keyPressEvent ignores modifiers, so the semantic behaviour is unchanged. - Document the dialog's selection and insertion rules at the top of the file. - No REQUIRE expectation values changed. Notes - No source files under test were modified; only the test tree and the test CMakeLists. - Mouse-coordinate tests in MkEdit_Tests.cpp (checkbox / link clicks) remain sensitive to widget geometry and should be run against a real platform plugin, not the "offscreen" one.
Enable test CMakeLists.txt(add)
Reset:settingsdialog.ui
upupwrite
marked this pull request as draft
September 24, 2026 17:28
upupwrite
marked this pull request as ready for review
September 24, 2026 17:28
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR mainly adds Linux support and Ctrl+A to select all text. I used an LLM to review most of the code, added Linux compatibility, and fixed a few small issues along the way. The comments and a lot of the more complicated code were generated by the LLM. I verified the main project source myself. The tests were also LLM-generated, and I haven't checked them line by line.
After fixing the header case issues, the project still doesn't run as expected on my machine. Basic text input doesn't even work, and it's pretty far from how it behaves on Windows. The refactor is fairly large, but GitHub Actions passes. I can't set up a proper Windows dev environment on my end, so I need someone to actually run it and verify it works on Windows, especially basic text input and Ctrl+A select all.
Thanks.