No RESUME. No session_id is retained for the main gateway and no sequence number is tracked, so every reconnect is a fresh op-2 IDENTIFY. It works — REST catch-up (catch_up_channel) covers the gap — but it is more expensive than resuming and loses anything catch-up does not reach.
Member lists are one 100-person window, SYNC only. The op-14 lazy-guild subscription requests range [[0,99]], and update_member_list applies only the SYNC operation — INSERT, UPDATE and DELETE are deliberately ignored (documented at discord.rs:3199-3207). So a roster goes stale as soon as anyone joins or leaves, and anything past 100 members is simply absent.
Done looks like:session_id and s retained, op 6 sent on reconnect with a fall back to IDENTIFY on an invalid-session reply; and the remaining member-list ops applied, with further ranges requested for large channels.
Filed from a protocol coverage survey of the four deployed protocols, 2026-08-29 (moho 1a11250, nobilis 030707f).
Two related gateway gaps.
**No RESUME.** No `session_id` is retained for the main gateway and no sequence number is tracked, so every reconnect is a fresh op-2 IDENTIFY. It works — REST catch-up (`catch_up_channel`) covers the gap — but it is more expensive than resuming and loses anything catch-up does not reach.
**Member lists are one 100-person window, SYNC only.** The op-14 lazy-guild subscription requests range `[[0,99]]`, and `update_member_list` applies only the `SYNC` operation — `INSERT`, `UPDATE` and `DELETE` are deliberately ignored (documented at `discord.rs:3199-3207`). So a roster goes stale as soon as anyone joins or leaves, and anything past 100 members is simply absent.
**Done looks like:** `session_id` and `s` retained, op 6 sent on reconnect with a fall back to IDENTIFY on an invalid-session reply; and the remaining member-list ops applied, with further ranges requested for large channels.
---
_Filed from a protocol coverage survey of the four deployed protocols, 2026-08-29 (moho `1a11250`, nobilis `030707f`)._
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Two related gateway gaps.
No RESUME. No
session_idis retained for the main gateway and no sequence number is tracked, so every reconnect is a fresh op-2 IDENTIFY. It works — REST catch-up (catch_up_channel) covers the gap — but it is more expensive than resuming and loses anything catch-up does not reach.Member lists are one 100-person window, SYNC only. The op-14 lazy-guild subscription requests range
[[0,99]], andupdate_member_listapplies only theSYNCoperation —INSERT,UPDATEandDELETEare deliberately ignored (documented atdiscord.rs:3199-3207). So a roster goes stale as soon as anyone joins or leaves, and anything past 100 members is simply absent.Done looks like:
session_idandsretained, op 6 sent on reconnect with a fall back to IDENTIFY on an invalid-session reply; and the remaining member-list ops applied, with further ranges requested for large channels.Filed from a protocol coverage survey of the four deployed protocols, 2026-08-29 (moho
1a11250, nobilis030707f).Finished — nobilis
d7a8ec0