Update Protos - #2
Open
github-actions[bot] wants to merge 1 commit into
Open
Conversation
github-actions
Bot
force-pushed
the
nightly-proto
branch
from
May 4, 2026 11:49
cbf4f2e to
864a3e2
Compare
github-actions
Bot
force-pushed
the
nightly-proto
branch
from
May 18, 2026 13:45
864a3e2 to
7bd19a5
Compare
github-actions
Bot
force-pushed
the
nightly-proto
branch
from
June 1, 2026 16:02
7bd19a5 to
2b8e259
Compare
github-actions
Bot
force-pushed
the
nightly-proto
branch
2 times, most recently
from
June 22, 2026 15:24
386d90f to
5d414a9
Compare
github-actions
Bot
force-pushed
the
nightly-proto
branch
2 times, most recently
from
July 20, 2026 12:22
4726f93 to
7490685
Compare
github-actions
Bot
force-pushed
the
nightly-proto
branch
from
July 27, 2026 13:01
7490685 to
cdb0925
Compare
insipx
pushed a commit
that referenced
this pull request
Aug 6, 2026
…koff, and catch-up ahead of enablement (xmtp#3887) Pre-enablement hardening for the XIP-83 bidi surface, from a fresh adversarial re-review of the now-merged transport + catch-up + lifecycle code (the interaction bugs that per-PR reviews miss). Bidi still ships **default-OFF** (`XMTP_BIDI_STREAMS_ENABLED`), so none of these block a release — they harden the fleet behavior that matters the moment the flag is flipped. Six atomic commits, one per finding. ## The two that matter most for enablement - **Order the lifecycle command with the suspend flag under one lock** (`#1`, medium). `suspend_bidi_streams`/`resume_bidi_streams` set the process `suspend_requested` flag under the registry lock but sent each transport's `Cmd::Suspend`/`Cmd::Resume` *after* dropping it. On a multi-threaded runtime a concurrent suspend+resume could race between lock-release and their sends, delivering the commands to a transport opposite to the flag — a wire left **live while backgrounded** (battery) or **parked while foregrounded** (no delivery). Now the synchronous command push (`enqueue_suspend`/`enqueue_resume`) happens inside the same locked section as the flag; only the reply await runs outside. - **Grow the reconnect backoff on a flapping wire** (`#2`, medium, pre-existing). The backoff reset to the 100 ms floor the instant a wire *opened*, before any frame. A server that accepts then immediately closes (an overloaded node RSTing after accept) pinned every client at the floor forever, amplifying load against the struggling backend. Now only a wire that stays up ≥ `MIN_STABLE_UPTIME` resets the backoff; one that dies inside that window is flapping and keeps backing off. The explicit `resume()` reset is unchanged. ## Latent hardening - **Guard the catch-up open against an empty subscription set** (`xmtp#3`, low). An empty `subs` would seed an adds-nothing open frame that never earns a `CatchUpComplete`, stalling ~90 s → `Exhausted`. Unreachable today (the welcome topic is always present), but the safety sat far from the open; short-circuit it to `Complete`. - **Time-box the catch-up half-close** (`xmtp#4`, low). `catch_up_attempt` awaited `conn.finish()` with no deadline; a wedged actor could hang it. Wrap the half-close and drain under one budget, matching the transport's own `close_gracefully`. - **Conclude a superseded resume waiter on suspend** (`xmtp#5`, low). A `resume()` awaiting catch-up that a `suspend()` preempts was left parked until some *later* resume, stranding the caller (and its task) for the whole background sojourn. A suspend supersedes pending resumes; conclude them at the suspend instead. ## Coverage - **Chunking × reconnect and × suspend/resume** (`xmtp#6`, test). The multi-wave-per-lease chunking across a wire death or suspend/resume was exercised only by the property test (16-case default); every scripted lifecycle test used the production cap, so none chunked. Adds two deterministic tests at a shrunk cap (over-cap lease survives a wire death; survives a suspend/resume), and raises the property test's default case count. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
github-actions
Bot
force-pushed
the
nightly-proto
branch
2 times, most recently
from
August 17, 2026 10:30
c441f9d to
d136d3b
Compare
github-actions
Bot
force-pushed
the
nightly-proto
branch
from
August 24, 2026 10:42
d136d3b to
a3d12b0
Compare
github-actions
Bot
force-pushed
the
nightly-proto
branch
from
September 7, 2026 15:37
a3d12b0 to
cb91a44
Compare
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
1 participant
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.
Auto-generated by create-pull-request