mirror of
https://github.com/penpot/penpot.git
synced 2026-11-15 22:54:45 +00:00
10 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
|
|
35b0a893e6 |
🚧 Save Phase 4 random harness and grid fixes in progress
Save Stage 2 structure, text, geometry and layout generation, independent oracle checks, frontend seeded replay and plugin visible-value checks. Include pending grid sync and deletion fixes with focused regressions. WIP validation: component sync 21 tests / 106 assertions; move shapes 6 tests / 33 assertions. Earlier plugin checks passed 68 full cases, 11 unit tests and types. The final generated replay, golden check, 5000-case gate, full suites and live WASM check remain unfinished. Stages 3 (swaps/variants) and 4 (library files) have not started. F124 benchmark ratios against Phase 1e: main 10/100/1000 copies 0.48/0.52/0.54; three levels 0.53; chain 0.56; library 0.52; many shapes 0.42; variant switch 0.64. No regression flag. The F125 deletion change still needs its final benchmark. Save the detailed resume checklist and logs in the local stabilization plan. Keep that workflow document out of the commit. AI-assisted-by: GPT-6 |
||
|
|
122f64db3d |
✨ Store a swap inside a copy as an override group
A swapped copy sub-head now keeps the shape-ref of the main child it replaces (its position), takes its content from its own component, and carries the touched group :swap-group. It used to point at the root of the new component and store a swap slot naming the replaced child. Every path reads and writes that form: - sync pairs a swapped head with the child it replaces and syncs it from its own component; it takes from the replaced child only its placement: position, constraints and layout item attributes; - reset undoes a swap in place, keeping the head's id and the other overrides of the copy; - detach, paste and duplicate make a head that leaves every copy level a plain copy of its own component, and advance a deeper one by position; the children of a swapped head never advance; - a variant switch keeps a swap stored this way, and counts only the copy's own swap, not one made in a main of the chain; - update main replaces the main child with a clone of the swapped one, and from a swapped head updates its own component; - a swap inside a copy keeps the old shape-ref and adds :swap-group; swapping back to the component of that position gives a plain nested copy whose children refer to the position's children; a swap inside a main is a plain nested copy. Files that hold swap slots keep working until a migration converts them. The component memories describe the model. This fixes what came from the slot: reset of a swapped child cleared the other overrides of the copy (F12) and gave copies of a main new ids (F66); update main left a swap mark (F16); a second variant switch lost the overrides (F29); grid cells did not pair a swapped child (F44); a swapped child did not follow its position (F65). F49 keeps failing, in another form. The tests that asserted slots now assert the override group; the tests of the slot code build the old form with :as-slot?. Benchmark (lower quantile, ms; baseline / now / ratio): deepest-edit-through-three-levels 35.2 / 35.4 / 1.01 edit-through-a-chain-of-dependents 68.9 / 68.7 / 1.00 library-update-into-copies 83.1 / 77.5 / 0.93 main-edit-into-10-copies 1.8 / 1.6 / 0.92 main-edit-into-100-copies 15.4 / 15.4 / 1.00 main-edit-into-1000-copies 153.1 / 151.1 / 0.99 variant-switch-into-copies 32.1 / 32.4 / 1.01 AI-assisted-by: claude-opus-5-5 |
||
|
|
270d0ac35d |
✨ Make the component test suites fail when their claims break
A review found tests that would pass with what they test broken. - Runners: both judge one variant at a time, fail a variant with no assertion and a known-failure mark that applies nowhere, and give the asserter the state the operations produced. The pure runner validates the final file and rejects operations of unknown kind. The frontend interpreter fails a step when the store reports an event error (Potok only logs it), reports asserter errors against their case, and stops the watchers that install! starts. - Nodes drive the real code: a main root rename goes through end-rename-shape, and the clipboard's advance of nested copies is one common function, cll/advance-copied-objects, that copy-selected and both test paths call. - Cases N, AH, AI, AK, AD, AE, K, C, H, S, AY, I, Y, AX and F check what they claim; O, R and T get real sizes. - Stricter common unit tests found F60 (a repair detaches healthy copies), F61 (a switch drops a fixed-sizing size), F62 (the ref chain never stops at its target) and F63 (latent); all pinned. - Frontend library tests: the RPC double records on subscribe, the dropped commits check that nothing syncs, and the link and share tests check the events sent. - Plugin suite: OpSwap and OpDetach throw when the API call did nothing, and case AB checks that the reset head is an instance of the original component. - The composable suite memory records the new rules. |
||
|
|
eee28e36bb |
♻️ Tidy the composable suite and make its cases stricter
A review of the suite after the new operations: - Dead helpers and duplicate nodes go; the lineage reset and swap nodes are constructors of reset and swap-shape. - Repeated helpers are shared: grid boards (tho/add-grid-frame, tho/place-in-grid-columns), n/nested-head-under, n/child-shape, r/sync-cascade (the benchmark uses it too), the frontend fixture and fill helpers, and the plugin getMainComponent. - Runner tests that repeated cases B, C and D go. - Comments say what the code does, and the wrong ones are fixed. - Case K's remote edit painted red on red and its unedited variants asserted nothing; it starts grey and every variant asserts. - F43 is pinned in one case, AJ, by the value one undo leaves. L, M, Q and K get real frame sizes, since their pins depended on when the 100 ms layout buffer flushed. - make-nested-component takes a level count and depth-sweep gives each depth once, so K, Q, R, T and W stop running depth 1 twice. - The completion-contract tests no longer depend on fixed pauses or on how the watcher batches its events. Benchmark against the baseline (lower quantile): ratios 0.92 to 1.03, no world flagged. |
||
|
|
2cae81f588 |
✨ Add operations and cases to the composable component suites
The composable runners could edit, add, remove and move children, swap and switch, but not run the other component operations. New nodes run each one through the production generator in the pure runner and the real event in the frontend interpreter: - Detach, UpdateMain, DeleteComponent and RestoreComponent; - Duplicate, and NestCopy (a copy moved into a main); - EditText, with the automatic rename the text editor applies; - MoveShape (frontend only, through the sidebar) and MoveToCell; - swap-shape and reset on any copy head; - Undo with a step count, and an edit undone before its sync (frontend only). New cases, all with the undo/redo round trip: - Q, R, S, T, U: detach, update main, delete and restore, duplicate and nest, at nesting depths 0 to 2; - V, W, X: a new edit after undo, three undos over nested cascades, an undo before the edit's sync; - Y: text overrides against main text edits; - Z: a main child move; - AA to AF: swap outcomes on values and structure, and the grid cell of a swapped child; - AG, AH, AI: the plugin suite's slot alignment cases. The plugin suite gets OpDetach, OpSwap and OpReset, names that keep the roles of two nestable components apart, Assert.knownFailure, and ports of cases Q, AA, AB and K's reset. Supporting changes: nodes can record what they created (IAppliedResult); create-component takes a size, since with 0.01 frames the frontend reflows split undo entries at points that depend on timing (F43); the interpreter ends the text measurement headless, so text steps settle; tm/known-failure pins the wrong value a known bug gives in a case assertion; the interpreter no longer moves its undo baseline after a skipped optional, which cut the end-of-variant round trip short. Confirmed and pinned: F12 (resetting a nested swap clears the top copy's overrides), F27 (the automatic text rename touches :name-group), F29 (a second switch loses an override) and F44 (a swapped grid child misses the main's cell changes). F43 also shows through a plain sync at depth 2 (case Q). |
||
|
|
a70712fce4
|
🐛 Fix issue with compononent not sync on move (#11981) | ||
|
|
806a9ee98e
|
⬆️ Update dependencies (#12150) | ||
|
|
36cf4f39e1
|
✨ Apply method Penpot.waitForLayoutUpdate, removing workarounds (#11433)
* MCP server: Update instruction manual to point to method instead of sleep * MCP plugin: Call waitForLayoutUpdate prior to every code execution * Composable test suite: Use waitForLayoutUpdate instead of 200ms delay in tests |
||
|
|
014ec34249
|
🐛 Fix system crash on grid element deletion of a component (#10956)
* 🐛 Add regression test: main-side component edits break copy swap slots Reproduces the :missing-slot referential-integrity failure ("Shape has been swapped, should have swap slot") that crashes files with component copies. Root cause: reordering or deleting a nested sub-head IN THE MAIN of a component, while copies exist, does not propagate swap slots to the copies. find-near-match matches a copy's sub-heads to the main's children by POSITION, so once the main's order changes the copies' shape-refs no longer match their position and, lacking a swap slot, fail referential-integrity validation. - Copy-side edits are handled correctly (characterization tests, pass today). - The two main-side tests fail today with :missing-slot and go green once the sync assigns swap slots to copies on a main reorder/delete. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> * 🐛 Fix integrity crashes from copy/main child-order divergence Copy sub-heads were matched to their main's children purely by position (find-near-match), while several code paths reorder or mutilate one side only. Any of them made file validation fail with :missing-slot ("Shape has been swapped, should have swap slot"), crashing the workspace on the next validated commit, or persisting a corrupt file whose later edits crash. Reproduced live: deleting a sub-head inside a copy (which only hides it) and then reflowing the copy's grid moved the hidden (cell-less) child to the front of :shapes, knocking every sibling out of its positional slot. Four fixes, one per divergence path: - validate/check-required-swap-slot: a sub-head whose shape-ref is still a child of the near main parent is a REORDER (the component sync realigns it), not a swap; a swap slot is required only when the ref points outside the near main parent (a real swap). This matches how the sync engine itself pairs children (by shape-ref, not by position). comp-processors/fix-missing-swap-slots (migration 0019) is aligned: adding slots to merely-reordered sub-heads would freeze them out of normal synchronization. - changes/:reorder-children now refuses to alter the child structure of component copies unless allow-altering-copies is set, mirroring the is-valid-move? rule of :mov-objects; that structure is owned by the component sync engine. Grid reflows emitted this change type with no guard. pcb/reorder-grid-children also skips copy grids producer-side. - layout/reorder-grid-children keeps children that participate in no cell (hidden or absolute positioned) at their original index instead of lumping them at the front: moving them gratuitously changed their z-order and, in copies, broke the positional matching. Note :shapes stays reversed relative to the sorted cell order for in-cell children. - logic/generate-delete-shapes: deleting shapes from inside a main (without deleting the main root, whose copies keep working against the deleted component) now also deletes the copy shapes that reference them, transitively (copies of copies) and across all pages of the file, so no dangling shape-refs remain. Skipped for allow-altering-copies flows (component swap replaces the shape and the sync reconciles copies via swap slots). Cross-page removals build redo/undo changes against that page's objects directly, since the changes-builder mounts only the current page; their undo mov-objects carry allow-altering-copies so restoring inside a copy is not rejected by the new guard. The regression tests assert the fixed semantics: main-side reorders and deletes keep copies valid, the previously crashing full chain (unvalidated main reorder + later copy edit) stays healthy, :reorder-children cannot scramble copies, and reorder-grid-children keeps cell-less children in place. The namespace is now also registered in the JS test runner. AI-assisted-by: Claude Opus 4.8 (1M context) * ✨ Extend composable slot cases to the grid-reflow crash sweep Case D (CopySubheadDeletePreservesSlots) now sweeps which copy sub-head is deleted (first or last) and whether the copy root is resized afterwards, forcing a grid reflow: the reflow used to move the hidden (cell-less) child to the front of the copy's children, shifting every sibling out of its positional slot. Deleting the FIRST sub-head masked the bug (moving it to the front is a no-op), which is why the case passed before this sweep. The foundation layout is grid accordingly. Case E (MainReorderKeepsCopySlots) no longer hangs the app now that a main-side reorder leaves a valid file, so its warning docstring is replaced: it runs as a routine test (verified headless, passing) and is safe in a "run all". SlotIntegrity's doc is updated to the new validator semantics (a slot is required only for real swaps, not reorders); the positional alignment it asserts remains the correct, stronger steady-state invariant for these cases. The lockfile change materializes the playwright devDependency already declared in package.json. AI-assisted-by: Claude Opus 4.8 (1M context) * 📚 Record copy/main order-divergence invariants in memories Swap-slot semantics (membership, not positional; slots only for real swaps), the copy-structure guards on :mov-objects/:reorder-children, the grid reorder stability for cell-less children, and the main-side delete propagation in generate-delete-shapes. AI-assisted-by: Claude Opus 4.8 (1M context) * 🐛 Add adjustements to the code --------- Co-authored-by: Michael Panchenko <michael.panchenko@oraios-ai.de> Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
|
caacd482af
|
✨ Add composable test framework plugin (applied to component tests, runs in CI) (#10679)
* ✨ Add plugin with composable test framework and component tests The plugin provides a framework for writing composable tests against the Plugin API, and applies it to systematic end-to-end testing of component semantics. The framework's core ideas: a test is written once as a composition of operations over a starting configuration; choice points among the operations (optional steps, alternatives) expand the composition into a full sweep of test variants, so a single case definition yields broad combinatorial coverage; and the operations drive the real Plugin API with real change propagation, testing the full production implementation. The initial application is a suite of component test cases covering synchronization, overrides, swap slots and variants — the TypeScript/e2e continuation of the ClojureScript composable test suite (frontend_tests.composable_tests). Several cases originate from reproducing real defects (e.g. #10109 and the swap-slot corruptions). Tests run from an interactive panel in Penpot: cases are listed with plain-language descriptions, tests can be run selectively, results stream in live, and every checkbox carries a stable DOM id — so the panel can equally be driven programmatically (the basis for running the suite in CI), as documented in the plugin's README. Lives at plugins/apps/composable-test-suite as a regular member of the plugins workspace (init script, start:plugin:composable-test-suite, shared dev port 4202, covered by build:plugins via the new ./apps/*-test-suite filter). Related to #10584. AI-assisted-by: claude-fable-5 * ✨ Run the composable test suite headlessly in CI Adds a headless run mode for the composable test suite, following the plugin-api-test-suite's CI architecture, and a workflow that runs it as a per-PR gate. An in-sandbox entry (src/ci/headless.ts) runs the suite without the panel UI — the framework's runner was UI-free by construction, so no refactoring was needed — and streams each result through console markers, addressed by the same composite identifiers the panel uses (e.g. MainEditSyncs-2), with durations and, on failure, the error and the applied-steps transcript. It is built as a single self-executing bundle and evaluated directly inside a real Penpot plugin sandbox by the driver (ci/run-ci.ts), so no plugin dev server or port is involved. The driver needs no backend and no login: it serves the prebuilt frontend bundle via the frontend e2e static server and intercepts every backend RPC with Playwright fixtures. The mocked backend is not a limitation for this suite — everything it asserts is frontend store logic executed in memory — which the full run confirms: all 48 tests behave identically to the interactive panel, including variants and swap slots, with the single (currently expected) failure of MainEditSyncs-2 reproducing bug #10109 under the mock. TEST_FILTER selects tests by identifier substring; CI_TIMEOUT_MS bounds the run. The mock harness mirrors the frontend e2e harness (see the provenance note in the driver). Related to #10584. AI-assisted-by: claude-fable-5 * 📚 Restructure the composable-tests memory around both suites Present the composable component tests top-down: the shared framework principles upfront, then the two implementations — the ClojureScript suite in the frontend test tree and the TypeScript suite in the plugin, which tests fully end-to-end with a slightly more elaborate set of abstractions — and the plugin's headless CI run, pointing to the plugin's README for operational details. Also records this session's additions (geometry operations, case N, the CI harness). AI-assisted-by: claude-fable-5 * 📎 Refine the PR-description conventions in the creating-prs memory Encourage digestible descriptions: bullet items over prose (grouped by area with bold lead-ins for larger PRs) and no manual line wraps, since the rendered markdown adapts to the viewport. Also drop the outdated 'MCP' from the standard Note line. AI-assisted-by: claude-fable-5 * 🔧 Set Prettier endOfLine to auto in plugins workspace Prettier defaults to endOfLine "lf", which is incompatible with checkouts on Windows that use core.autocrlf=true * 🐛 Fix problems with suite --------- Co-authored-by: alonso.torres <alonso.torres@kaleidos.net> |