fix(ios): move a Pocket card to the top of its screen instead of sliding up a copy by valentinfernandez1 · Pull Request #1423 · paritytech/trinity-user-agents

valentinfernandez1 · 2026-10-14T18:28:09Z

Fixes paritytech/platform-bugs#96.

What

Tapping a card in the Pocket moves that card to the top of the screen, and the product fades in below it. Closing moves the card back to its place in the Pocket. There is one card on screen the whole time, as on Android.

The Pocket hides its own copy of the card while the card is open.
If the Pocket cannot show the card when the animation starts, for example because the product switched to the Chat tab, the whole screen fades instead.
A card opened from a link, or with Reduce Motion on, uses the plain full-screen cover.
A card published with faceShown: false lands with its face shown, then folds it away, as on Android.

Screen.Recording.2026-10-15.at.9.32.42.AM.mov

How

The card's screen animates its own presentation (PocketCardFlight). The card moves while the rest of the screen fades, with Android's timing: 500 ms, the same overshoot curve.
The Wallet tells the screen where each card is and hides the open one (PocketCardOrigin).
ModuleNavigating.presentFullScreen takes the transitioning delegate.

Tests

PocketCardFlightTests checks that the Pocket's copy is hidden while the card is open and comes back after any close, and that the screen and the reused product end up as they do without the animation.
The published face test in PocketCardScreenTests now covers an open with and without the animation.
Each test was checked by breaking the code it covers. The full app test target passes locally.
On the iPhone 16 simulator (iOS 26): opening and closing the last card and a card in the middle of the stack, and closing with the face folded away.

The Widget under an opened Pocket card can show or hide the face above it through a new widget-only ExpandedCard trait (wire id 23), and a card definition's faceShown sets where the face starts. The host sizes the Widget to the space the face leaves and never locks the face against the user.

…dapter

…ate test RUNTIME.md now lists ExpandedCardHost with the other platform traits and says what a connection without one answers. The placeholder test from the protocol PR duplicated expanded_card_without_an_adapter_is_unsupported; its reason moves into that test's doc. The runtime borrows the adapter instead of cloning it.

…ocket test host setup RUNTIME.md said Platform combines every listed trait except the optional ones, which included ExpandedCardHost by omission; it now says the hook travels per connection and is in neither set. The trait doc no longer compares it with PermissionStatusHost, which is installed process-wide. ExpandedCardFaceOutcome derives SCALE Encode/Decode like HostContactPick, matching the codec codegen publishes for it. The Pocket and expanded card test hosts share one setup.

…host

…its card

…s faceShown: false

A debug product with no published app showed "Couldn't load this product." under its Pocket card. The product form now takes an App URL on 127.0.0.1; the card's page loads from it with the card query appended, keeping the product's identity through the session, the permission client and the bridge bootstrap, which can no longer read it from the host. The form also gains an "Open with the face away" switch for debug cards.

…ed card has settled The fold cleared its dragging flag before the release animation ran, so a request in that window was answered applied and then dropped by the animation lock.

…, and document the expanded card dev loop Clears the stale opening value when another card is selected, corrects the opening-fold comment, and describes the App URL and face-away fields, the live expanded card loop and faceShown in the Android dev-loop doc and the sample README.

…wiring touched

Host the expanded product only for the card still selected, withdraw a face request's reply when the core stops waiting so the fold does not move after a NotPresented answer, clear a deleted product's debug App URL, and make faceShown explicit at every PocketCardDefinition and DebugPocketCard construction.

…view A card's WebView is served from the product's widget executable on the base-domain origin, as iOS does, and the card warm-up fetches that same archive; a product with no widget falls back to its base host rather than running its app as a Widget. The opening-face lookup waits up to five seconds, since the page waits on the same product resolution anyway. A card fading out no longer answers the next card's face requests. The session starter derives the execution kind from the card it is given, and the debug App URL is refused unless it is a loopback http url.

…ts to its screen

…hat the card hosts return

…e page below it The card screen now answers setFaceShown for real: it scrolls the face away or back, answers UserMoving while the user drags or the face still glides after a drag, and keeps a request made before the first layout for that layout. A card created with faceShown false opens with the face away. The page is sized to the screen the face leaves it, so its bottom edge stays in view, and is resized only when the face comes to rest. While the face moves the page reaches the bottom of the screen, so no blank strip opens under it. The scroll's content is a screenful plus the face, independent of the page.

…carries its face A page move that a drag cuts short no longer fits the page mid-drag, and the page grows only for an animated move, since a move made at once is fitted straight away. The tests now catch the page being resized on every scroll frame, cover a tap that stops a gliding face, and wait for animated moves by polling for the expected state instead of a fixed delay.

A card in a worker manifest may carry "faceShown": false to open onto its page with the face out of the way. Left out or null, the card opens with its face, and a value that is not a boolean fails the pocket section's decode, so the product loses its cards and keeps its worker, as Android reads the same manifest. Opening a card now looks the card up in its product's published manifest first, for at most 500 ms; a lookup that fails or runs out of time opens the card with its face shown.

Opening a card now waits on its product's manifest before presenting, and a second tap or deeplink in that window built a second screen, which claimed the product's surface and page from the first. PocketCardHosts now ignores an open while another is under way. The 500 ms bound on that lookup did not hold against the real resolver: a task group waits for its cancelled child, and ProductResolver waits on a shared unstructured task that cancellation does not reach. The lookup and the deadline now race through a continuation that resumes once, so the card opens with its face on time whatever the resolver does. A parser test also pins that one card with a non-boolean faceShown drops the valid cards beside it, as Android does.

…at overlap Every test that shows a card's screen now does so in a window of the app's scene, through one fixture that presents it as the app does and one that closes it. The screen's sizing tests become one, a surface test whose second half could not fail on its own goes, and the card hosts' surface checks join the tests of the products they come with.

One lookup that answers late unless cancelled shows both that the face stays shown and that the lookup is told to stop. It also covers the card a product does not publish, whose lookup fails the same way, so that test and the slow resolver go.

The debug card's face switch is checked where the menu saves it, and a page execution's kind in one comparison over both pages.

Drop screen tests that pinned internal mechanics of page fitting and repeated requests, and the bridge test that passed without the override. Merge the two "published face ignored once moved" tests into one parameterised test, and the five surface tests into one test that walks a card from never shown, through presenting, a never-shown rival screen and a full-screen cover, to closing.

A bad address in a debug-only field shows up as the card's error screen, which is enough. Fold the card id and product casing lookups into the widget URL composition test, and drop the menu save tests.

…the code

…ing up a copy Tapping a Pocket card presented a full-screen cover that slid up with its own copy of the card, so two cards were on screen. The card's screen now animates its own presentation: its card moves from the card's place in the Pocket to the top, and back on close, while the rest of the screen fades. The Pocket hides its copy while the card is open. Same timing and curve as Android (500 ms, cubic-bezier 0.34, 1.2, 0.64, 1). A card the Pocket cannot place (opened from a link, removed, or the Wallet tab not showing) and Reduce Motion keep the plain cover. Fixes paritytech/platform-bugs#96

…os' into fix/pocket-card-open-transition-ios # Conflicts: # hosts/ios/polkadot-app/Modules/Wallet/WalletMain/WalletMainProtocols.swift

github-actions · 2026-10-14T18:28:59Z

This pull request touches an app, which is not built by default. Add a label for each build you want:

iOS simulator build, ios-simulator-build: builds the iOS app, runs its tests and attaches a build for the iOS Simulator on a Mac.
iOS device build, ios-device-build: attaches a signed build that installs on a registered iPhone or iPad.

Each starts as soon as it is added and follows the branch from then on.

github-actions · 2026-10-14T18:36:25Z

Bundle size report

Compared with main at 679ca10.

	Raw	Gzip	Brotli
`truapi-host`	7.89 MiB (-4.6 KiB, -0.1%)	5.66 MiB (-1.6 KiB)	5.39 MiB (-890 B)
`truapi-provider`	4.03 MiB (+2.7 KiB, +0.1%)	1.27 MiB (+142 B)	960.9 KiB (+790 B, +0.1%)
`truapi`	6.30 MiB	891.5 KiB	730.0 KiB
Total	18.22 MiB (-1.9 KiB)	7.80 MiB (-1.4 KiB)	7.04 MiB (-100 B)

WebAssembly modules

	Raw	Gzip	Brotli
`truapi-host/wasm/web/truapi_server_bg.wasm`	2.78 MiB (-4.6 KiB, -0.2%)	990.8 KiB (-1.6 KiB, -0.2%)	746.9 KiB (-1.0 KiB, -0.1%)
`truapi-host/wasm/web/truapi_verifiable_bg.wasm`	4.89 MiB (-13 B)	4.64 MiB (+60 B)	4.61 MiB (+129 B)
`truapi-provider/truapi_provider_bg.wasm`	3.99 MiB (+2.7 KiB, +0.1%)	1.26 MiB (+142 B)	953.1 KiB (+785 B, +0.1%)

Changed files (3)

	Raw	Gzip	Brotli
`truapi-host/wasm/web/truapi_server_bg.wasm`	2.78 MiB (-4.6 KiB, -0.2%)	990.8 KiB (-1.6 KiB, -0.2%)	746.9 KiB (-1.0 KiB, -0.1%)
`truapi-provider/truapi_provider_bg.wasm`	3.99 MiB (+2.7 KiB, +0.1%)	1.26 MiB (+142 B)	953.1 KiB (+785 B, +0.1%)
`truapi-host/wasm/web/truapi_verifiable_bg.wasm`	4.89 MiB (-13 B)	4.64 MiB (+60 B)	4.61 MiB (+129 B)

_{Commit: 504f691}

github-actions · 2026-10-14T18:37:46Z

CI Status: 24 required jobs green, 15 passed and 9 skipped by path filter.

All job results

job	result
android-bindings	skipped
bundle-size	success
changes	success
changeset-guard	skipped
cli-package	skipped
codegen	success
e2e	skipped
explorer	success
headless-install	success
host-android-bindings	skipped
host-android-detekt	skipped
host-wasm	success
ios-bindings	success
ios-swift	skipped
licenses	success
playground	success
provider-android-bindings	skipped
release-guard	success
rust	success
ts-client	success
ts-debugger	success
ts-host	success
wasm-provider	success
workflow-lint	skipped

Signing credentials: failure as of 2026-10-15, a release may fail

_{Commit 504f6912 · run log}

The screen now hides its card in the Pocket when it is presented and shows it again when it is dismissed, so opening and closing agree even when only one of them flies. The flight is handed the slot it was decided on instead of looking it up again, and every path through it ends the flight. Two flight tests are merged into one, and the one that pinned the old hiding rule is gone.

A product that opens a chat closes its card and switches tabs at once, so the slot looked up when the close was asked for no longer belonged to the tab the close returns to. The flight now looks the slot up when it starts and fades the whole screen when there is none, and the Wallet answers by the tab container's selection, which changes before the old tab is taken off. A close also reports the landing fade in its duration. The published-face test now covers both an open that flies and one that does not, in place of a separate flight test.

sergkononenko · 2026-10-15T13:10:51Z

UI/UX behavior note: user should be able to pull card down to dismiss the details and return back to Pocket.

What I would like to see sine we are changin/adding a new transition:

Generic card details screen, so the Balance and ID cards open the same way and the in-place expansion is gone
The cross-fade from the collapsed card to the expanded one
Pull to dismiss
The cards after the opened one sliding off (as they do now)

…card-face-ios # Conflicts: # hosts/ios/polkadot-appTests/TrUAPI/RustRuntimeBridgeTests.swift

…os' into fix/pocket-card-open-transition-ios

The typed address is parsed once when the card is saved, so the card holds a URL rather than a string parsed later. An empty field still means no page, since an empty string parses to none.

…card-face-ios

…d-open-transition-ios

…transition-ios # Conflicts: # hosts/ios/polkadot-app/Modules/Products/Pocket/PocketCardOpening.swift # hosts/ios/polkadot-app/Modules/Products/Pocket/PocketCardScreenViewController.swift # hosts/ios/polkadot-appTests/Products/PocketCardScreenFixtures.swift # hosts/ios/polkadot-appTests/Products/PocketCardScreenTests.swift

valentinfernandez1 added 30 commits October 7, 2026 18:44

feat(truapi): add the ExpandedCard trait, reachable from a Widget

51adaf1

fix(truapi): answer a host without expanded cards with Unsupported

cc62ded

feat(truapi): answer a face request made mid-drag with UserMoving

829ede1

feat(truapi): route the expanded card face to a per-connection host a…

988381e

…dapter

feat(truapi): ask the native host to move an expanded card's face

a6d14b8

feat(ios-sdk): let a host move an expanded card's face

03ff539

feat(android-sdk): let a host move an expanded card's face

b3ae3fe

chore(changeset): the host runtime answers expandedCard on a browser …

d8658b8

…host

feat(android): run a Pocket card's page as a Widget and let it reach …

496db3c

…its card

fix(android): bound a card face request by one timeout

8f0d4af

feat(android): let a Pocket card's page fold the card

8591e4d

feat(android): open a Pocket card with its face away when it publishe…

d4c8a3b

…s faceShown: false

fix(android): read a null faceShown as absent, as iOS does

bae2b92

chore(sample): an expanded card page that moves the Pocket face

ebc2362

chore(sample): make the expanded card page readable in recordings

7c23734

fix(android): answer a page's face request user moving until a releas…

74db699

…ed card has settled The fold cleared its dragging flag before the release animation ran, so a request in that window was answered applied and then dropped by the animation lock.

style(android): keep the original import order in the files the card …

02c0951

…wiring touched

feat(ios): run a Pocket card's page as a widget and route face reques…

2cc1dfe

…ts to its screen

test(ios): prove a card's face request reaches its screen, and name w…

b4b5882

…hat the card hosts return

valentinfernandez1 and others added 9 commits October 9, 2026 11:39

test(ios): fold single-check Pocket card tests into their neighbours

287bc61

The debug card's face switch is checked where the menu saves it, and a page execution's kind in one comparison over both pages.

refactor(ios): stop checking the debug Pocket card widget URL

41e4d22

A bad address in a debug-only field shows up as the card's error screen, which is enough. Fold the card id and product casing lookups into the widget URL composition test, and drop the menu save tests.

docs(ios): trim the expanded card dev loop and comments that restate …

ad317c6

…the code

Merge branch 'main' into feat/pocket-expanded-card-face-ios

02ad023

Merge remote-tracking branch 'origin/feat/pocket-expanded-card-face-i…

46e1219

…os' into fix/pocket-card-open-transition-ios # Conflicts: # hosts/ios/polkadot-app/Modules/Wallet/WalletMain/WalletMainProtocols.swift

valentinfernandez1 added the host-ios Touches the iOS host tree label Oct 14, 2026

valentinfernandez1 added 2 commits October 14, 2026 15:53

valentinfernandez1 marked this pull request as ready for review October 15, 2026 12:02

valentinfernandez1 requested a review from a team as a code owner October 15, 2026 12:02

valentinfernandez1 added this pull request to stack #1441 October 15, 2026 12:22

valentyna-kozlova requested a review from alexander-lisovyk October 15, 2026 12:42

alexander-lisovyk approved these changes Oct 15, 2026

View reviewed changes

valentinfernandez1 added 2 commits October 15, 2026 11:20

Merge remote-tracking branch 'origin/main' into feat/pocket-expanded-…

1cf40b2

…card-face-ios # Conflicts: # hosts/ios/polkadot-appTests/TrUAPI/RustRuntimeBridgeTests.swift

Merge remote-tracking branch 'origin/feat/pocket-expanded-card-face-i…

87e47b8

…os' into fix/pocket-card-open-transition-ios

pgherveou approved these changes Oct 15, 2026

View reviewed changes

valentinfernandez1 added 3 commits October 15, 2026 14:07

refactor(ios): store a debug card's widget page as a URL

13afc92

The typed address is parsed once when the card is saved, so the card holds a URL rather than a string parsed later. An empty field still means no page, since an empty string parses to none.

Merge remote-tracking branch 'origin/main' into feat/pocket-expanded-…

28691b2

…card-face-ios

Merge branch 'feat/pocket-expanded-card-face-ios' into fix/pocket-car…

dce82e6

…d-open-transition-ios

Base automatically changed from feat/pocket-expanded-card-face-ios to main October 15, 2026 17:35

Navigation Menu

Repository navigation

fix(ios): move a Pocket card to the top of its screen instead of sliding up a copy - #1423

fix(ios): move a Pocket card to the top of its screen instead of sliding up a copy#1423
valentinfernandez1 wants to merge 89 commits into
mainfrom
fix/pocket-card-open-transition-ios

valentinfernandez1 commented Oct 14, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 14, 2026

Uh oh!

github-actions Bot commented Oct 14, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Oct 14, 2026 •

edited

Loading

Uh oh!

sergkononenko commented Oct 15, 2026 •

edited

Loading

Uh oh!

This branch has not been deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

Conversation

valentinfernandez1 commented Oct 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What

How

Tests

Uh oh!

github-actions Bot commented Oct 14, 2026

Uh oh!

github-actions Bot commented Oct 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Bundle size report

Uh oh!

github-actions Bot commented Oct 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sergkononenko commented Oct 15, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This branch has not been deployed

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

valentinfernandez1 commented Oct 14, 2026 •

edited

Loading

github-actions Bot commented Oct 14, 2026 •

edited

Loading

github-actions Bot commented Oct 14, 2026 •

edited

Loading

sergkononenko commented Oct 15, 2026 •

edited

Loading