Discovery transcript: GE-BUG-055

Revision as of 14:23, 18 October 2026 by CodexBot (talk | contribs) (Add broadcast filter discovery excerpt)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

Discovery of GE-BUG-055

Related bug: Classic bug: Filtered broadcast suppresses later recipients
Source: Discord #dev discussion, September 3, 2026. Times are as displayed in the supplied screenshot; their time zone is not established.
Editorial note: This is a short excerpt from one screenshot. The bug article gives the source-level explanation and distinguishes classic's single filter from ge-next's category-based settings.

September 3: A hail disappears downstream

10
50 PM — manicpoporg
wow, i found an interesting legacy bug
if message filters (set filter on in classic, or the category-based filters in ge-next) filter out a broadcast message for one user, it also filters it out for all users below that one
let's say you have users 0, 1, 2 and 3 and user 2 has hails set to off. if user 1 hails user 0 will see it but not 2 (correct) or 3 (incorrect)

[At 11:07 PM, GitHub announced commit 4beca4d, whose displayed title begins “remove obsolete mbbsemu mitigations, fix prf/fi...”.]

Clarification

“Below that one” was informal wording. The example shows that the sender's hail reached slot 0, was intentionally suppressed at slot 2, and was then incorrectly absent for slot 3. The affected recipients are those processed after the filtered slot in the broadcast loop, not lower-numbered slots. Source review showed that rejecting one recipient cleared the shared print buffer used by the remaining recipients. ge-next 0.3.2 changed delivery so each recipient's filter affects only that recipient.

See also