Feat/scroll scenes - #1060
Merged
Merged
Feat/scroll scenes#1060
Conversation
The generated TqMotion's scroll trigger animated ONLY the y axis (one useTransform into style.y), silently dropping x/scale/rotate/opacity/filter from the from/to keyframes — while the editor canvas interpolated all of them. Same UIDL, two behaviors. The scroll path now mirrors the canvas runtime: progress (windowed by scrollOffset, optionally useSpring-smoothed by scrub) writes the FULL interpolated state straight to the DOM node via useMotionValueEvent — zero React re-renders per frame, blur() interpolation included. End-to-end test asserts the parity markers in the generated wrapper. First slice of the feat/scroll-scenes initiative (P0). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…tes (P1) New TqScrollScene widget: a tall track with a sticky pinned stage whose scroll progress (useScroll, optionally useSpring-smoothed) drives data-scroll-bind property lanes on descendants — multi-keyframe, compositor- only props (x/y/scale/rotate/opacity/blur), named presets (depth-1/2/3, fade-window, rail-x, zoom-through, unblur-in) with JSON as the escape hatch. Lanes are applied by DOM query with a MutationObserver so runtime Repeater items are picked up — never React child introspection. Reduced-motion settles on final or static per scene config; --scene-progress exposed only when configured; dev warning when an ancestor's overflow disables sticky. Element mapping scroll-scene-node → TqScrollScene; framer-motion reused. End-to-end test covers page emission, binding passthrough and the wrapper's guardrail markers. Lane semantics mirror the GUI renderer's scroll-scene-lanes module verbatim. Part of feat/scroll-scenes P1. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…editor Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…pective Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
pop-in, tilt-in, blur-up, pulse and float — same states, verbatim, as the canvas renderer's motion-keyframes, so the published site matches the editor for every new preset. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Mirrors the canvas renderer: exact heights let content taller than one screen be painted over by the following section. Regression-guarded in the end2end test. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Mirrors the canvas renderer: travelling lanes no longer spill a scrollbar onto the published page, and scene stacking is contained. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
layout='chapters' makes the stage a centering grid and same-cell-stacks every direct child (scoped child rule) — mirrors the canvas renderer. End2end guarded. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…pers Mirrors the canvas: the CSS child rule stays for plain children, and a restack pass pins the real grid items inline on mount and on observer ticks, so wrapped/late-rendered content stacks too. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…afe style tag - The style tag emits via dangerouslySetInnerHTML (React SSR escapes text children of style, and style is a raw-text element that never decodes entities — the selector broke server-side and hydration failed on the text mismatch). - The stacking pass mirrors the editor runtime: pierces open shadow roots and slots, re-runs on customElements.whenDefined for not-yet-upgraded hosts, self-heals stale grid-area off contents wrappers, and is fully duck-typed (no instanceof). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…ve watcher The clean-then-rewrite regeneration deleted every generated directory and streamed the tree back while the generated project's next dev was watching it. That fed the watcher seconds of half-written state (transient compile errors that sometimes wedged the dev server until restart), and webpack drops its watch on a deleted-then-recreated directory — after which hot reload silently stops and re-syncs appear to do nothing. Generation now packs into a staging directory and reconciles into the project dir: byte-identical files produce no watch event at all, orphans are still deleted file-by-file (the stale-orphan guarantee that motivated the clean), live directories are never deleted, and PRESERVE_ON_CLEAN is honored at the root exactly as before. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…pper The scroll-linked parity work replaced useTransform with useMotionValueEvent + useSpring in the emitted framer-motion import; this suite still asserted the old line and failed only on CI's full run — the targeted suites run during development never included it. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- pure-DOM TqScrollVideo: scene discovery via data-scene-track, window start/end mapping, exp smoothing, reduced-motion poster/end, seek replay on loadedmetadata; the wrapper owns the size (absolute video + contain:layout that survives an authored position:static) - styled-jsx plugin emits widget-element classes as :global(...) — styled-jsx stamps its scope hash only on host DOM elements, so every class handed to a Tq* React component compiled into a rule that matched nothing and widgets silently dropped their authored styles (a Scroll Video shipped with zero height) - scene track stamps data-scene-track; TqScrollScene + TqScrollVideo complete the five-registry ceremony (mapping, widgets, component-path plugin, rules-of-hooks + plugin tests, end-to-end suite) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Brings in v0.43.54 and the any-AI-provider preparation for the assistant (#1062) — provider utils across the workflow AI nodes, the ai-chat route and provider generator, plus development's in-place stagger rewrite of the motion widget. git called this merge clean, and it was not. Only one file is touched by both sides, widgets/motion-component.ts, and both sides independently added the SAME two helpers to the generated runtime — ours near EASINGS for the scroll-linked interpolation, theirs above cssFromState for the in-place cascade. Different regions, so git kept both, and the emitted component declared TRANSFORM_KEYS and transformComponent twice: every generated project with a motion node would have died at parse time with "Identifier 'TRANSFORM_KEYS' has already been declared". The bodies were byte-identical; one copy survives, carrying both comments' reasoning, and both features build on it. Proof: the generated runtime parses (babel, jsx) at 513 lines with no duplicate top-level const, and carries our scroll path (applyScrollState, the tilt-in/blur-up presets) alongside their cascade (cssFromState). 2774 tests pass across 269 suites after a full lerna build. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Brings in v0.43.55 and the e-commerce fixes (#1063): email rendering, cart resurrection after checkout, and product-variant availability. Zero conflicts and zero file overlap with the scroll work — their changes are confined to the e-commerce and workflow-node generators. Verified beyond the suite, since a clean merge is not proof for template-string generators: the motion (513 lines), scroll-video (202) and scroll-scene (425) runtimes all parse with no duplicate top-level declarations, and so do the workflow shared (1306) and client (516) runtimes, whose executor-generator they edited. 2822 tests across 272 suites pass after a full lerna build. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
- The generated TqScrollVideo wrapper no longer forces an inline position:relative — inline beats the CSS class carrying the backdrop's absolute-fill, which collapsed the video to a 0px in-flow div (in the JSX, URL live, nothing on screen). contain:layout alone keeps the wrapper the containing block, mirroring the canvas renderer. Regression test pins the emitted wrapper style. - Scroll scene default layout is chapters; viewport-pass scrubbing is a legacy fallback (contract comments). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
An absolutely positioned clip paints above static in-flow siblings, so a plain heading dropped next to the backdrop video vanished behind it. TqScrollScene now always injects a backdrop rule pinning any direct scroll-video child into the stage's negative z band (isolation on the stage keeps it inside the scene), targeting a new data-scroll-video marker stamped on the TqScrollVideo wrapper — scene children layer above the clip however the video arrived. Mirrors the canvas renderer. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…w (Next controller + HTML CSS) - Scene/Motion widgets write translate/scale/rotate, never transform; scene stage gets perspective; overscroll-behavior-y none on pages with a scene and html background synced from body (no white band). - Scene backdrop rule pins the scroll video under the chapters. - Next: NextSnapIntoViewProjectPlugin emits components/tq-snap-into-view.js (proximity CSS + one rest-triggered settle, gentle/firm reach, nearest-wins, reduced-motion aware) and injects it into _app only when an element opts in. - HTML: pluginSnapIntoView appends the native proximity rule to the global stylesheet (inline in each page when there is none), registered in packProject's HTML branch. - e2e for both, absent-when-unused asserted. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…ript) - Next: NextScrollRailProjectPlugin emits components/tq-scroll-rail.js — rail CSS (item snapping gentle/firm, hidden scrollbar) and the wheel driving (one item per gesture on a snapping rail, free otherwise, hands back to the page at either edge) — injected into _app only when a rail opted in. - HTML: pluginScrollRail appends the rail CSS to the global stylesheet and, when a rail asks for wheel driving, one inline script per page; shared global-css helper (appendGlobalCss / appendPageScript) also used by the Snap-into-view plugin; registered in packProject's HTML branch. - e2e for both, absent-when-unused asserted. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…ydration - uidl-resolver: data-scroll-*/data-snap-*/data-chapter-* stay on the element that has a box instead of moving to the (display:contents) link wrapper — a linked chapter element ignored its scroll lanes. - uidl-resolver: editor bookkeeping (data-scroll-bind-rel, data-chapter-window, data-chapter-template) is dropped from every generator's output. - next: body custom code ships inert in <template data-tq-body-code> and the TqBodyCode _app sibling unpacks it after hydration (scripts recreated in document order; noscript stays live) — the slider bundle no longer rewrites slides before React hydrates, which reported every slide as a mismatch. - next: custom-property element styles (--swiper-*) proven to pass through. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…er rule Mirrors the editor's lane contract: clip-path insets for the reveal lanes, a rounded --tq-count variable plus the [data-scroll-count]::before counter rule for the count lane. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
data-scene-hidden attribute + stage CSS rule (mirrors the editor's lane contract) instead of inline pointer-events, so authored pointer-events survive; e2e asserts the attribute ships. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…soft cap Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Chapters stack inside the sticky stage, so a native hash jump to a chapter's id lands at the track top, never at the chapter. The scene now translates hash navigation (link clicks, hashchange, arriving with a hash) into track progress — the midpoint of the chapter's story window (data-chapter-window, else its lane stops) — and scrolls there smoothly. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…geometry chapterSnap="gentle": when scrolling pauses mid-scene, progress settles smoothly on the nearest chapter's moment. Never intercepts an in-flight scroll; reduced-motion visitors are never moved; a scrubbed background video glides to the chapter's frame. Chapter-window helpers are now shared between hash navigation and the snap. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…ment TqScrollScene aimed its gentle snap and anchor navigation at the midpoint of a chapter's window — the last beat still rising. The runtime now derives each chapter's SETTLED MOMENT (every lane arrived, none leaving) from the root's and descendants' lanes, honours the author's data-chapter-moment pin first, falls back to the window's middle only for pure through-motion, includes chapters whose root has no lanes, and settles in the direction the visitor was travelling. settledMomentForLanes lives in scroll-scene-moment.ts as plain JS whose source is injected into the generated runtime — the same function the editor draws on its timeline (mirrored in teleport-gui). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Development locked the emitted-component roster at 18; this branch's tq-scroll-scene and tq-scroll-video rows make it 20. Both pass the analyser. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…tions - NextPageTransitionProjectPlugin wraps the page mount in _app with a generated TqPageTransition (AnimatePresence mode wait, frozen exit at the scroll offset, reduced-motion passthrough, first load never animates). - Direction-aware: back button plays the reverse preset, Circle grows from the last pointer press; cover family paints overlay panels outside the keyed page. - Skipped routes appear instantly (arrival-only); [param] segments match any value. - Preset options resolved at generation time (defaults filled, unknown words replaced); cover colour is a project custom property, a literal colour, or the brand token chain — nothing else reaches the generated file. - UIDL types + validator decoder carry globals.settings.pageTransition (preset, duration, easing, skipRoutes, options). Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…ions - A `custom` preset carries the arriving-start and leaving-end states (opacity, x, y, scale, blur) in the UIDL; duration and easing travel in their existing fields. Values are clamped and the easing validated at generation time (mirrored from the editor), so a hand-edited file never breaks a build; the generated wrapper plays them through the same variants. - UIDL types and the validator decoder accept `custom`. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…xt listener The published TqScrollScene tracks the chapter on stage (shared chapter enumerator for pinned and un-pinned scenes, reduced-motion visitors included), stamps data-chapter-active / data-chapter-count and dispatches tq-chapter-reached for every chapter passed, in travel order, via a microtask; replaced nodes keep the stamp, effect re-runs re-announce. The workflow plugin emits a listener bound to the chapter root that ignores nested scenes' events, honours once, and reads the stamp when it attaches late. activeChapterIndex picks the latest moment passed. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…e, Next listeners The published TqScrollScene announces every point (quarter, half, three-quarters, end) the visitor scrolls DOWN past as a bubbling `tq-scene-point-passed` event on its track, one microtask after the progress change, in scroll order, and stamps the furthest point passed as `data-scene-point` together with the dispatch so a listener attaching in the same effect flush gets the event and a later one reads the stamp. The first measure starts below the scene, so a page that opens already past a point (anchor link, restored scroll) counts it. Going up announces nothing. The workflows plugin gains two lifecycle triggers: `event-scene-scrolled-past` (bound to the scene, resolved to its track, nested-scene bubbles ignored, `once` detaches, late attach fires from the stamp) and `event-page-scrolled-past` (window scroll progress over the scrollable height, non-scrollable page = 1, rAF-throttled scroll + resize, checked on attach, re-armed above the point). The scene trigger is pruned per page like element-visible. `SCROLL_POINT_PROGRESS` is the one map both sides pin. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Plays, pauses, play-or-pauses, restarts, mutes or unmutes the bound Video element (or the first <video> inside the bound element). A play the browser blocks for sound is retried muted and reported as `mutedFallback`; blocked even muted reports `autoplay-blocked`; a missing video reports `video-not-found` instead of throwing. Registered as client-only. Written without object spread so the ES5 build never pulls a helper into the handler that ships by toString. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…n, full-width chapters A scene's direct media or data-scene-backdrop child renders behind the chapters and is excluded from chapter enumeration; ancestors with overflow hidden are unclipped at mount so the pin holds; the motion wrapper isolates its paint order; chapters take the full stage width. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
…ainer becomes the anchor - reserveProjectStyleNames: the project style sheet is global, so a generated element class that equals one of its classes inherits it. A page named "Manifesto" keyed its root `manifesto-container` while another page had authored a `.manifesto-container` grid, and the whole page became that grid. The project's class names are reserved before keys are generated; a colliding element takes a numbered key through the existing counter. - A container that is nothing but a link (no style, class, binding, event or runtime attribute) becomes the anchor itself instead of `<a><div>…</div></a>`. Imported markup writes `<a><img></a>` and stylesheets size the image by its place in the tree (`.surface-media > a > img`); the extra <div> broke that chain and the image fell back to its natural size. An element with a box of its own keeps the wrapper. - scroll-video end-to-end test: the backdrop rule now lists several selectors. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
… link itself - Motion in the static HTML export: scroll scenes, element motion (load, in view, stagger in place, hover, tap, scroll-linked), scroll video and page transitions (cross-document View Transitions: the back button plays the preset reversed, a circle grows from the press point, cover and curtain sweep the chosen colour, a page that opted out appears at once). One runtime file with only the parts a project uses; no JavaScript, a blocked runtime and reduced motion all fall back to readable content. - One engine for both exports: the plain-JS parts of TqScrollScene, TqMotion, TqScrollVideo and the page-transition registry live in teleport-shared as source text. The emitted Next components are byte-identical. - Links: a plain div that is a flex or grid item becomes the anchor itself. A display: contents wrapper cannot take keyboard focus in Chrome; on a generated store 124 of 124 links are now reachable (was 24), with no size change. Wrappers that remain carry data-thq-link-wrapper for the editor's focus-ring rule. Every link rule is pinned by a test. - teleport-test/scripts: real-browser checks for the HTML export and its page transitions, a Next-versus-HTML comparison and a link-shape measurement. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
… fixed; reveals fit the screen - On Visible: one settings reader and one observer for both generator paths. The editor stores the threshold as a percent and the run choice as triggerOnce; the generated IntersectionObserver threw on any typed threshold and "Trigger once" was never read. Percent -> fraction, once ?? triggerOnce, fires on becoming visible. - HTML download: state conditions resolve against the page state's starting value (the shipped generator dropped every state-conditioned element), and EXPORT_PLAYBACK in teleport-types tells the editor what each export plays. - A picture flies between pages, in both downloads: following a link that holds a picture to a page showing the same picture flies it onto its twin while the page plays its transition. Stylesheet and pairing helpers live in teleport-shared; Next plays that one change as a same-document View Transition. pageTransition.flyingPictures: false (UIDL type + validator) turns it off in both. - Scroll video: the whole clip is held in memory on published sites, so a seek no longer waits on the network; "Play back to the start" maps the scroll to the clip there and back. - Page transitions (Next): a reveal (Circle, the Wipes) is measured on the screen, not on the page. On a long page it covered the screen almost at once and the new page popped in; leaving a scrolled page it closed above the screen. Its clip never stays on the page, the first page included (a clip left there broke the fixed header and pinned scenes while scrolling); it runs frame by frame (no blink as it ends); the arriving page waits up to 300 ms for the pictures on its first screen. - HTML download: an element's own styles weigh two classes (.name.name), as in the Next export and the editor, so its own colour beats a two-class combinator there too. - teleport-test/scripts: Next browser checks (the flight, reveals on tall pages), one test site for both exports, a clip-seek measurement. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
…eamed by the second - Page transitions: Circle and the Wipes open the arriving page OVER the leaving page, which holds still underneath (view-transition-css.ts plan.revealsOver). Played one after the other, the leaving page was wiped away to the site's bare background first — a white sheet between every two pages. The HTML download keeps the old picture unanimated and plays the arrival with no delay (back: tq-page-arrive-back on the arriving picture only). The Next export plays every reveal navigation as a same-document View Transition where the browser has them (startViewChange, HoldPage, the arriving page pictured once its first-screen pictures are decoded); without them the two pages overlap (AnimatePresence "sync"), the arriving page laid above the frozen one until its animation completes. Pixel proof on a production copy: the share of the screen showing neither page fell from 0.82 to 0.00. - Scroll video engine (teleport-shared scroll-video-engine-source, one text for the Next export and the HTML download): holdClip reads the fragmented encode's index from its first 64 KB and, where MediaSource supports the codec, streams the seconds around the scroll position (streamClip: the second on screen first, then ahead, then behind, out to ±12 s); whole in memory where it cannot, as before. A 12 s clip is scrubbable after 315 ms instead of 1,712 ms, downloading 1.03× the clip instead of 1.25×. - teleport-test html-motion scripts: verify-transitions (21 checks), verify-next-transitions (16), measure-scrub, README. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
vladgrecu
marked this pull request as ready for review
September 24, 2026 17:38
Two conflicts, both sides kept: the workflow component plugin imports resolveScrollPoint and findFormStateKeyForHydration; teleport-shared exports ScrollSceneRuntime, MotionRuntime, PageTransition and WebPush. Development's link-prop-case test now expects the boxless link wrapper's marker attribute. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… unit glide on the scroll-linked path
- TqMotion (Next): the load trigger is driven by the "played" flag set on the
first frame after mount (animate: played ? toVars : fromVars) instead of
framer's initial -> animate pair. A page transition wraps every page in an
AnimatePresence with initial={false}, which skips the mount animation of
every motion component under it: an entrance never played on the first
page a visitor opened, and a loop (repeat -1) never started and stood at
its end state.
- Shared motion engine: a value with a unit (0% -> -50%, 2rem) is
interpolated in that unit when both ends share it, as the canvas does.
- Tests: motion-end2end (the load flag), teleport-shared motion-engine (4).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…trances are never left invisible; strict HTML keeps its runtime - teleport-shared: storyLanes, clearMovement and letRowsSwipe in the scene engine, letMarqueeSwipe in the motion engine, run by the Next widgets and the HTML runtime alike. - Next, TqMotion: the still element for less motion is drawn only once the page is mounted. The server renders the animated element with its starting style (opacity 0) and React keeps markup it disagrees with, so every entrance stayed invisible for those visitors (published since 0.43.22). TqPageTransition reads less motion when a page change starts instead of rendering the bare page (React error 418 on every load). - HTML: insertBeforeClosingTag (global-css.ts). With strict HTML whitespace Prettier writes the head's closing tag across two lines, so the motion runtime was never linked, the page-transition script never injected and the CSS fallback landed before the doctype. The runtime link, the page-transition script and both global-css helpers now find the tag in either form. - Tests: story-lanes (5), motion-engine (+2), global-css (2, new), the HTML motion and page-transition strict-way tests, the Next motion and page-transition tests (one each, three assertions updated). - Browser proof on a generated HTML and Next site, with and without reduced motion: 46 of 46 checks; 0 blocks left hidden, 0 hydration errors. Full suite: Test Suites 514 passed, 514 total; Tests 1 skipped, 5930 passed, 5931 total; tslint 0. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…y one small script A downloaded HTML site now behaves like the Next export for state: a phone menu opens, a price switch flips, tabs switch their panel and move their active class. Every branch of a condition is in the page; the ones a visitor does not start on carry `hidden`, so the first paint, a visitor without JavaScript and search engines see the page as it first loads. The generator writes readable bindings (`data-tq-if`, `data-tq-on-click`, `data-tq-class`, and `data-tq-scope` / `data-tq-state` on the page root and on each component instance, which keeps its own state) and ships `tq-state.js` beside the stylesheet, linked only from pages that use it, the way `tq-motion.js` is. The script reads the bindings with a small parser and never evaluates code, so it runs under a strict Content-Security-Policy. Click workflows that only set local state, such as the editor's Tabs, compile to the same bindings. Shapes it cannot express keep today's first-load rendering. `EXPORT_PLAYBACK` now says HTML runs conditions, which lets the editor keep them in the UIDL and drop its download warning. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Conflicts, both sides kept: - next generator utils: the base's inline script error guard beside the branch's body code wrapper. - uidl-resolver abilities: the base's data-source link pass beside the branch's box-less link wrapper mark. One behaviour git could not see: the base's guard knows the project's own scripts by the document's address, and the branch runs body custom code after hydration, as scripts inserted later. Chromium reports no file for an error in such a script, so the guard let it through (measured over http with the built packages: 2 of 2 errors reached the app's own listener; WebKit names the document and stopped both). The unpacker now names the document as the source of every inline script it recreates; data blocks (JSON-LD, templates) and external scripts stay as written. With it 0 of 2 reach the app in Chromium and in WebKit. Whole suite on the merged code: 526 suites, 6022 tests passed, 1 skipped. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
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.
No description provided.