iris is the framework alone; the app is one crate in app-rust/

Iris: "the organization of the rust rewrite is a mess right now... there
shouldn't be anything related to the app inside of iris. Iris is supposed
to be the UI framework alone." And, on the crate count: "I'm confused why
the app only code needs more than one crate though."

Nine cargo workspaces become three, and the port's project code -- which
sat in five places, four of them inside the framework -- becomes one crate,
`ai-app`, in `app-rust/`:

  client-core                -> app-rust/src/client
  iris/transcript-ui         -> app-rust/src/ui
  iris/transcript-fixture    -> app-rust/src/ui/fixture.rs + tests/ + touch/
  iris/desktop-app           -> app-rust/src/desktop + src/bin_desktop.rs
  iris/android-app           -> app-rust/src/android + android-project/
  android-shell              -> app-rust/src/shell

iris/ keeps core, macro, the iris crate, tabs-ui and rig-input, and now
mentions no session, transcript, setup or server anywhere.

Only two of the old splits had a reason that survived reading. event-model
stays a crate at the repo root because server/ depends on it too, so a
crate is what makes the backend and the app agree by construction. The two
Android .so names looked like a hard constraint -- a package produces one
library artifact -- until P2 turned out to already plan merging those two
Android apps into one; both faces now come out of libai_app.so, picked
apart by features so `--no-default-features --features shell` keeps wgpu,
parley and iris out of the Compose app's APK. docs/RUST.md's "One app
crate" has the rest, including what each remaining feature is for.

DECISIONS.md and SUBAGENTS.md move into docs/ with everything else.

Verified: ./run-tests.sh and `cd iris && cargo test` green, clippy and fmt
clean in all five workspaces, `cargo ndk -t x86_64` links libai_app.so,
build-apk.sh produces an APK that installs and launches on this checkout's
emulator (Gl ... virgl, as expected), and the phone-sized headless
screenshot renders the transcript unchanged.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
This commit is contained in:
irisandClaude Opus 5 committed 2026-09-08 23:36:38 -04:00
1 parent e9a6562dc6
commit 6d5a231f5c
100 files changed
+924 -3295

No files matched your search

+270
View File
@@ -0,0 +1,270 @@
//! I5's desktop proof: the transcript screen built from synthetic
//! `ai_app::client::transcript_fold` rows (no network, no server -- see
//! `lib.rs`'s doc for why `transcript-ui` itself never fetches anything),
//! run via `iris/run-headless.sh transcript -- -p transcript-ui` for a
//! screenshot on the winit backend, or `cargo run --example transcript -p
//! transcript-ui` with a real compositor.
//!
//! The rows exercise every one of the seven "hard to get back" behaviours
//! this box's markdown/selection work is meant to show: a heading, bold,
//! italic, an inline code span, a link, a fenced code block (rich inline
//! text), a multi-message conversation (bottom-anchored virtualised list),
//! and a three-call tool run (collapsed by default -- tap it, or drive it
//! with `ui-trace record --do "tap 'Tools'"` on Android, to prove
//! hold-the-edge expand).
use ai_app::client::QuestionOption;
use ai_app::client::transcript_fold::{QuestionCard, TranscriptItem, TranscriptRow as FoldedRow};
use iris::prelude::*;
fn main() {
DefaultApp::<Client>::run();
}
#[derive(DefaultUiState)]
pub struct Client {
ui_state: DefaultUiState,
#[allow(dead_code)]
screen: ai_app::ui::TranscriptScreen,
}
fn msg(seq: u64, from_user: bool, text: &str) -> FoldedRow {
FoldedRow::Single(if from_user {
TranscriptItem::UserMsg {
seq,
text: text.to_string(),
attachments: Vec::new(),
}
} else {
TranscriptItem::AssistantMsg {
seq,
text: text.to_string(),
settled: true,
}
})
}
/// One tool call. `result` is `None` for a call with no result yet and
/// `Some((output, failed))` for one that answered.
fn tool_call(id: &str, tool: &str, input: &str, result: Option<(&str, bool)>) -> TranscriptItem {
tool_call_in("run1", id, tool, input, result)
}
/// The same, in a named run. Two runs in one transcript must not share a
/// `run_id`: it is the row's identity in the list (`row::row_key`), and
/// two rows under one key is the duplicate-key fault AGENTS.md's
/// "Importing" section describes. Here it made two rows swap cached
/// heights and draw at each other's boxes.
fn tool_call_in(
run: &str,
id: &str,
tool: &str,
input: &str,
result: Option<(&str, bool)>,
) -> TranscriptItem {
TranscriptItem::ToolRun {
seq: 3,
id: id.into(),
run_id: run.into(),
tool: tool.into(),
input: input.into(),
output: result.map(|(out, _)| out.to_string()).unwrap_or_default(),
done: result.is_some(),
failed: result.is_some_and(|(_, failed)| failed),
asks: Vec::new(),
images: Vec::new(),
}
}
/// A call stopped on the reader: one unanswered permission question.
fn asking(id: &str, tool: &str, input: &str) -> TranscriptItem {
let mut call = tool_call_in("run2", id, tool, input, None);
if let TranscriptItem::ToolRun { asks, .. } = &mut call {
asks.push(QuestionCard {
seq: 9,
id: format!("{id}-q"),
prompt: "Allow this command?".into(),
header: None,
options: vec![
QuestionOption {
label: "Allow".into(),
description: None,
preview: None,
},
QuestionOption {
label: "Deny".into(),
description: None,
preview: None,
},
],
multi_select: false,
answers: Vec::new(),
});
}
call
}
/// Longer than the card's own cap, so the "Show all N lines" control is on
/// screen in the expanded shot.
fn long_output() -> String {
(0..200)
.map(|i| format!("test ai_app::ui::case_{i} ... ok"))
.collect::<Vec<_>>()
.join("\n")
}
fn synthetic_rows() -> Vec<FoldedRow> {
vec![
msg(
1,
true,
"Can you show me a **bold** word, some *italic* text, and `inline code`?",
),
msg(
2,
false,
"# Sure\n\nHere's a [link to the repo](https://example.com/ai-app-2) and a fenced block:\n\n```rust\nfn main() {\n println!(\"hi\");\n}\n```",
),
// Every state a tool card has to draw, in one run (P1b): a call
// that worked, one the tool reported as failed, one whose result
// never arrived, and one still running. The last two look the same
// in the events -- an empty output and `done: false` -- and are
// told apart only by whether the session is still working, which
// is what `TranscriptScreen::set_session_working` says.
FoldedRow::Tools(vec![
tool_call(
"t1",
"Read",
r#"{"file_path": "src/main.rs"}"#,
Some(("fn main() {}\n", false)),
),
tool_call(
"t2",
"Bash",
r#"{"command": "cargo build --release", "timeout": 480000, "description": "Build it"}"#,
Some((
"error: could not compile `iris`\nCaused by: linker not found",
true,
)),
),
tool_call("t3", "Grep", r#"{"pattern": "fn fold_event"}"#, None),
]),
// A lone call is a card too rather than a group of one -- and this
// one carries the kilobyte output a collapsed card must not lay
// out.
FoldedRow::Single(tool_call(
"t5",
"Bash",
r#"{"command": "cargo test -p transcript-ui -- --nocapture"}"#,
Some((&long_output(), false)),
)),
msg(6, true, "Looks good, thanks!"),
// Every block kind `ai_app::client::markdown_blocks` names, in one
// row, so P1a's appearance can be looked at against the Compose
// app's without a server (docs/RUST.md's P1a box). The heading,
// paragraph, fence and table are the *same source* the bench
// fixture carries (`app/bench-fixture/generate.py`), so the two
// screenshots differ only in the renderer; the list and the quote
// are extra, because the fixture has neither.
msg(7, false, BLOCK_SAMPLER),
]
}
/// One of each markdown block, for the P1a screenshot pair. See
/// [`synthetic_rows`].
const BLOCK_SAMPLER: &str = "\
## What changed
Iris **fold** render measure session window anchor context transcript \
iris measure iris scroll call transcript layout *cursor* context, and a \
[bench](https://example.com/bench) link.
```rust
fn fold_event(items: Vec<Item>, seq: u64) -> Vec<Item> {
// a comment worth keeping: this is the fold the app's own screen runs
let mut out = items;
out.push(Item::new(seq));
out
}
```
| column | value |
|---|---|
| a | measure place draw tool call token context window anchor |
- one bullet
- another, with `inline code`
- nested one level
1. first numbered
2. second numbered
> A quoted line, to show the bar and the indent.
";
impl DefaultAppState for Client {
fn new(
mut ui_state: DefaultUiState,
rsc: &mut DefaultRsc<Self>,
_: Proxy<Self::Event>,
) -> Self {
let screen = ai_app::ui::build(rsc, &mut ui_state, synthetic_rows());
// Exercises `push_row`/`ItemKey` beyond construction time, matching
// how a live SSE loop appends -- a row arriving after the screen
// already exists must land at the bottom without disturbing what's
// above it (I3's `push_back`/`snap_end`).
screen.push_row(
rsc,
&FoldedRow::Single(TranscriptItem::CommandRow {
seq: 8,
text: "clear".into(),
}),
);
// A second run at the live end, so the *running* state is on
// screen too. It cannot share a row with "no result": the two are
// the same events and are told apart only by whether the session
// is working, which is a property of the row rather than of the
// call (`TranscriptScreen::set_session_working`).
screen.push_row(
rsc,
&FoldedRow::Tools(vec![
tool_call_in(
"run2",
"t6",
"Read",
r#"{"file_path": "docs/RUST.md"}"#,
Some(("# Moving the app to Rust\n", false)),
),
tool_call_in(
"run2",
"t7",
"Bash",
r#"{"command": "cargo clippy --workspace --all-targets"}"#,
Some(("error: unused variable `x`", true)),
),
tool_call_in("run2", "t8", "Glob", r#"{"pattern": "**/*.rs"}"#, None),
// Waiting on a permission, so this card is drawn *open*
// whatever the reader last chose -- the command is the
// thing being decided, and a row saying only "Bash"
// cannot be decided on. It is also how the expanded card
// (input block, output block, timeout) gets into the
// screenshot without a finger.
asking(
"t9",
"Bash",
r#"{"command": "rm -rf target", "timeout": 120000, "description": "Clear the build"}"#,
),
]),
);
screen.set_session_working(rsc, true);
// The expanded picture has no other way to be looked at on a
// machine with no display and no finger -- see `run-headless.sh`
// and docs/RUST.md's P1b box.
if std::env::var_os("IRIS_TOOLS_EXPANDED").is_some() {
assert!(
screen.expand_tail_tools(rsc, true),
"the newest row must be the tool run this flag is about"
);
}
Self { ui_state, screen }
}
}