Conversation
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 8aa847c7fc
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| auth_username: None, | ||
| auth_password: None, | ||
| auth_token: auth_token.map(|t| t.to_string()), |
There was a problem hiding this comment.
Pass username with token during gateway bootstrap connect
The deploy bootstrap path now always connects with auth_username: None while still passing auth_token, which means the handshake sends no auth credentials in this commit’s Connection::handshake logic unless a username is set. In the new kernel (gateway-os/src/kernel/connect.ts), token auth explicitly requires a username, so infra deploy/upgrade cannot apply runtime config to auth-protected gateways and will repeatedly fail to connect in that scenario.
Useful? React with 👍 / 👎.
| } | ||
|
|
||
| function canonicalizeWhatsAppActorId(jid: string | null | undefined, senderPn?: string): string | null { | ||
| const senderPnMatch = typeof senderPn === "string" ? senderPn.match(/^(\\d+)@/) : null; |
There was a problem hiding this comment.
Fix escaped digit regex in WhatsApp actor canonicalization
This regex uses \\d inside a regex literal, so it matches a backslash plus d instead of digits. As a result, numeric WhatsApp JIDs/senderPn values are never normalized to wa:+<number> and fall back to wa:jid:..., which breaks phone-based actor identity normalization (the same escaping issue appears again at line 143).
Useful? React with 👍 / 👎.
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: c21c2d1006
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| export class GatewayEntrypoint | ||
| extends WorkerEntrypoint<Env> | ||
| implements GatewayAdapterInterface | ||
| { | ||
| async serviceFrame(frame: Frame): Promise<Frame | null> { |
There was a problem hiding this comment.
Preserve legacy channelInbound RPC entrypoints
This change replaces GatewayEntrypoint with a serviceFrame-only interface, but the test channel in the same commit still calls env.GATEWAY.channelInbound(...) (for example in channels/test/src/index.ts). That mismatch is not type-checked across workers and will fail at runtime, so inbound test-channel traffic and related dev/e2e flows break immediately unless the channel is migrated in the same change or compatibility methods are kept.
Useful? React with 👍 / 👎.
| if (request.method !== "GET" && request.method !== "HEAD") { | ||
| init.body = await request.arrayBuffer(); | ||
| } |
There was a problem hiding this comment.
Stream git proxy request bodies instead of buffering
buildGitProxyRequest reads the entire request body into memory with await request.arrayBuffer() for non-GET/HEAD calls. Git push requests can contain large packfiles, so this introduces size-dependent failures under Worker memory limits and adds avoidable memory duplication before forwarding to ripgit. Forwarding the original stream (request.body) avoids this regression for larger repositories.
Useful? React with 👍 / 👎.
No description provided.