Skip to content

Add AssumeNeverEmits, RaceWithSignalAndTimer, and FromEventBuffered - #1

Merged
kblok merged 3 commits into
mainfrom
add-rx-primitives
Jul 27, 2026
Merged

Add AssumeNeverEmits, RaceWithSignalAndTimer, and FromEventBuffered#1
kblok merged 3 commits into
mainfrom
add-rx-primitives

Conversation

@kblok

@kblok kblok commented Jul 27, 2026

Copy link
Copy Markdown
Member

Motivated by puppeteer-sharp's WaitForTargetAsync/WaitForFrameAsync — all three promote patterns that already existed as private, duplicated workarounds in consumer code.

AssumeNeverEmits — promotes the NeverReached helper RetryAndRaceWithSignalAndTimer already had privately: widens an error-only Observable<Unit> to any T, since C# has no bottom type for RaceWith to accept the way TypeScript's Observable<never> does. RetryAndRaceWithSignalAndTimer now uses the public version instead of its own copy.

RaceWithSignalAndTimer — the non-retrying half of RetryAndRaceWithSignalAndTimer, for a single wait (e.g. "wait for the next matching event") that doesn't need retrying. RetryAndRaceWithSignalAndTimer is now built on top of it (Retry().RaceWithSignalAndTimer()).

FromEventBuffered — an eagerly-attaching, buffered variant of FromEvent. Ordinary FromEvent is cold — nothing attaches to the underlying event until Subscribe is called. That's fine in upstream rxjs, where single-threaded run-to-completion semantics mean nothing can fire between attaching a handler and checking some existing state for an already-matching item. .NET has no such guarantee (event delivery can run on another thread), so that gap is a real, provable race — FromEventBuffered attaches immediately and buffers into a ReplaySubject so nothing fired in that gap is lost. (This closes a real bug found while porting puppeteer-sharp's WaitForTargetAsync: a plain Subject bridge dropped a matching TargetCreated event that fired in exactly that gap, hanging every page creation until timeout.)

Also: consolidated every Extras class (the three above plus the pre-existing CancellationExtras/TimeoutExtras/FilterAsyncExtras/RetryAndRaceWithSignalAndTimerExtras) into a single public static partial class Extensions (in the existing RxSharp.Extras namespace), split across the same per-concern files via partial. Extension methods (FilterAsync, RaceWithSignalAndTimer, ...) never needed the class name typed out — using RxSharp.Extras; already surfaces them via dot-completion — but the three factory-style methods (no natural this Observable<T> receiver) gave callers no way to guess which class held which method. One name fixes that for zero cost to the extension methods.

All 839 tests pass (825 existing + 14 new).

🤖 Generated with Claude Code

https://claude.ai/code/session_01K92eapPm8e7mX7puT4cF4s

kblok and others added 3 commits July 27, 2026 15:30
Three additions motivated by puppeteer-sharp's WaitForTargetAsync/WaitForFrameAsync,
promoting patterns that already existed as private, duplicated workarounds:

- AssumeNeverEmits: promotes the NeverReached helper RetryAndRaceWithSignalAndTimer
  already had privately (widening an error-only Observable<Unit> to any T, since C#
  has no bottom type for RaceWith to accept the way TypeScript's Observable<never>
  does). RetryAndRaceWithSignalAndTimer now uses the public version instead of its own
  copy.

- RaceWithSignalAndTimer: the non-retrying half of RetryAndRaceWithSignalAndTimer, for
  a single wait (e.g. "wait for the next matching event") that doesn't need retrying.
  RetryAndRaceWithSignalAndTimer is now built on top of it (Retry().RaceWithSignalAndTimer()).

- FromEventBuffered: an eagerly-attaching, buffered variant of FromEvent. Ordinary
  FromEvent is cold - nothing attaches to the underlying event until Subscribe is
  called. That's fine in upstream rxjs, where single-threaded run-to-completion
  semantics mean nothing can fire between attaching a handler and checking some
  existing state for an already-matching item. .NET has no such guarantee (event
  delivery can run on another thread), so that gap is a real, provable race -
  FromEventBuffered attaches immediately and buffers into a ReplaySubject so nothing
  fired in that gap is lost.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K92eapPm8e7mX7puT4cF4s
Every Extras method previously lived in its own single-method static class
(CancellationExtras, TimeoutExtras, FilterAsyncExtras, ...), each named after its one
method - fine for the extension methods (FilterAsync, RaceWithSignalAndTimer, etc.),
which already show up via dot-completion on Observable<T> with just `using
RxSharp.Extras;`, but a real discoverability tax on the three factory-style methods
(FromCancellationToken, Timeout, FromEventBuffered) that have no natural `this
Observable<T>` receiver to hang off of: a caller had no way to guess which class held
which method.

All seven are now one `public static partial class PuppeteerExtras`, split across the
same per-concern files as before via `partial`. (Named PuppeteerExtras rather than
plain Extras: CA1724 flags a type named the same as its containing namespace, and
every file's own doc comments already frame these as puppeteer-sharp's real
motivation.) Two colliding private `DefaultCause` helpers (one per branch's default
exception factory) got distinct names now that they're members of the same type.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K92eapPm8e7mX7puT4cF4s
PuppeteerExtras tied the name too closely to one consumer for what's meant to be a
general-purpose combinator surface. Extensions in the existing RxSharp.Extras
namespace reads the same way as e.g. Observable in the RxSharp namespace.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K92eapPm8e7mX7puT4cF4s
@kblok
kblok merged commit 3d81516 into main Jul 27, 2026
1 check passed
kblok added a commit to hardkoded/puppeteer-sharp that referenced this pull request Jul 27, 2026
Bumps to RxSharp 0.1.2 (pending publish - see hardkoded/ReactiveExtensions-Sharp#1),
which adds three primitives that let these two methods read much closer to upstream's
merge(...).pipe(filterAsync(predicate), raceWith(...)) shape instead of the
ReplaySubject + manual predicate-in-the-handler version from the previous commit:

- FromEventBuffered eagerly attaches the raw event handler (instead of a plain
  Subject fed by a raw handler) and exposes it as a proper Observable, so predicate
  filtering can happen declaratively downstream via .Filter(predicate) instead of
  inside the handler.
- RaceWithSignalAndTimer replaces the hand-built cancellation/timeout race branches
  in WaitForTargetAsync.
- The private NeverReached helper duplicated in both Browser.cs and CdpPage.cs is
  gone, replaced by RxSharp's own AssumeNeverEmits.

Same external behavior as before - same exception types/messages. Verified with 8
repeated reruns of the two tests that first exposed the ReplaySubject race, confirming
FromEventBuffered doesn't reintroduce it.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K92eapPm8e7mX7puT4cF4s
kblok added a commit to hardkoded/puppeteer-sharp that referenced this pull request Jul 27, 2026
RxSharp's FromEventBufferedExtras/TimeoutExtras/CancellationExtras/etc are now one
public static partial class PuppeteerExtras - see hardkoded/ReactiveExtensions-Sharp#1
for why. Mechanical follow-up: this repo's two call sites.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K92eapPm8e7mX7puT4cF4s
kblok added a commit to hardkoded/puppeteer-sharp that referenced this pull request Jul 27, 2026
PuppeteerExtras -> Extensions, see hardkoded/ReactiveExtensions-Sharp#1.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01K92eapPm8e7mX7puT4cF4s
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant