Repository navigation
Send a DM on what its own messages say, not the Mac's chat label - #165
Conversation
The Mac still labelled one RCS contact's chat as iMessage, so Blip sent every message to them as iMessage, and every one failed with error 22. Another contact's chat was labelled SMS, so they got plain SMS, which the Mac logged as error 4 (Not Delivered). The iPhone delivered those anyway: each row had the contact's read receipt 1 to 6 seconds later. Found on a macOS 27.2 gateway, 2026-11-12. `imsg chats` reports the chat row's service_name. With one row per conversation, that label follows the Mac's own last send, Blip's included. So letting it override the window's evidence loops. nixfred#97, nixfred#147 and nixfred#151 each guarded one direction of that loop. The rule "only when the newest row is inbound" did not hold either, because the label is not the newest row's service. - mergeChats keeps sendServiceForMessages' answer for every DM the window has rows for. The label names only quiet DMs. - thread.ts reports `service` from the opened conversation's own rows. BlipView prefers it for text and file sends, through the new SendState.sendServiceArg (built into SendState.mjs). - After a failed iMessage to a phone, the next send goes the way they last texted us when that was RCS or SMS. Otherwise it is SMS, as before. - sendFailed(): a read receipt outranks message.error. This applies to the bubble, the receipt line, the failure toast and the service rules. Invariant added to CLAUDE.md. All 892 tests pass. Each change was checked by mutation: undoing it makes its tests fail. Not yet run: a live send to an RCS contact through the Mac's RCS service. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
Hi Jason, and thank you: this is a careful first PR. The diagnosis is right (the chat row's Checked and fine: merges cleanly on main Blocker: a merged phone + email conversation can now send SMS to an email address, and never recovers.
Fix: decide Should-fix (reproduced by the reviewer, I did not re-run these):
Nits: CLAUDE.md says "On macOS 26 and later, one row holds the whole conversation" but the evidence is one 27.2 gateway ("observed on 27.2" is safer); the test "a genuinely green DM still takes its service from the list" now passes trivially (window and list are both SMS), so rename or drop it; the new Ping me when you push; I will re-check the two fixtures first. Generated by Claude Code |
…review) Blocker: a merged phone + email conversation could send SMS to the email, and never recover. loadThread decided the service from every row in the cluster, but the send goes to the canonical id. The mirror case sent iMessage to an Android number after they last wrote from the email. - sendServiceForMessages(msgs, prefer, target) uses only rows under `target`. thread.ts asks threadSendService(), which returns no answer when the target has no rows of its own, so the label stands instead of "iMessage" (should-fix 4). foldThreadAliases keeps the canonical's own service, never an alias's. mergeChats falls back to the label when that is "". - Guard: an address that is not a phone number never gets SMS, in the rule and in SendState.sendServiceArg. It gets RCS only after texting us on RCS, as RCS business agents (name@rbm.goog) do. Their chats send RCS on main today. - Should-fix 1: if the newest of our sends that provably arrived (delivered or read) is an iMessage newer than their last text, the DM is iMessage. That keeps a quiet blue DM blue when their last text is an old SMS. An error-22 row is never delivered, so the nixfred#151 loop stays closed. - Should-fix 3: a read receipt clears the error only on SMS/RCS rows in a 1:1. An iMessage row or a group row keeps the Mac's verdict. - Should-fix 2: the collector's `recent` rows are not --rich, so the failure toast still fires for a relayed SMS. The comments now say so. - Nits: the CLAUDE.md wording is "observed on macOS 27.2". The test that passed trivially is replaced by the label fallback. BlipView's wiring moved into noteThreadService/sendServiceFor, and scripts/test-send-service-ui.py runs those shipping functions under QtTest against the real SendState.mjs (added to CI). Ignoring which chat the rows were loaded for now fails there. All 901 tests pass, and so do the four QtTest scripts. qmllint reports nothing new. Eleven mutations each fail a test: the target filter (rule, thread, fold), the email guard (rule, SendState, view), the label fallback, the receipt scope, the proven-iMessage rule, the empty-load answer, and ignoring the loaded chat. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
@nixfred I pushed Blocker (merged phone + email). Confirmed:
Guard against sending SMS to an address. A handle that isn't a phone number never gets SMS, in the rule and in Should-fix 1. A new rule sits between
It needs Should-fix 2. The code comments and the PR text now say it plainly: the collector's Should-fix 3. Should-fix 4. If no rows sit under the target, there is no Should-fix 5 / live data. These are from the reporting machine, metadata only. (Corrected: the first version of this paragraph converted the install time with the wrong time zone.)
Your check on a real Android contact is the one that counts. Nits.
Verified.
🤖 Generated with Claude Code |
|
A correction to my comment above. I converted the install time with the wrong time zone. The fix went in at 00:41 UTC on 2026-11-13, so the three RCS sends I cited (00:31–00:36 UTC) came before it and are not evidence for this PR. No send to either RCS contact has gone out since the install, so a Blip send through the Mac's RCS service is still unconfirmed. I've fixed the comment and the PR description. The rest stands: 240 sends since the install, and none has an error. |
|
Thank you Jason, this is exactly the right response to the review, and the correction about the install time is appreciated. Checked on 3a13a27: One thing blocks the merge now, and it is not your fault: #160 landed on main a few minutes ago ( After the rebase I will re-run the two fixtures and the mutation spot-check, and do the one thing only I can: a real send to an Android RCS contact on the 27.2 gateway (with Fred's go-ahead), since that is still the only open question on this PR. Generated by Claude Code |
…phic, X-POSTS row) WHAT: docs/x-post-2026-11-14.txt (5,465 characters, 14 handles each tagged once, no em dashes), the 1600x900 graphic docs/img/x-post-2026-11-14.png with its HTML source, and the X-POSTS.md row. WHY: Fred asked for a detailed thank-you post covering every contributor since the last posted update (Post 4, 26 Sep). Posts 5, 6, 7 and W45 were never posted, so this one replaces them: 40 merged PRs, 38 from 18 outside people, 2 ours, one cherry-pick (David Benedic), one reporter (edmooredop), and two PRs in review (Jason Alexander #165, Curtis Chambers #167). HANDLES: from CONTRIBUTORS-X.md (all 13 match); @jasona verified on his GitHub profile and x user; no X for Marcus (credit by GitHub handle, haplab.com), Erik, Gerald, Colton, Dima, David, edmooredop, Curtis: named without @. PLACED on vic's clipboard (image first so it sits in history, text last) and the PNG at ~/Pictures/blip-thank-you-2026-11-14.png on vic. LR-T: blip, x-post, contributors, thank-you, graphic LR-D: blip, docs, comms LR-K: blip-post-8-all-contributors-since-26-sep Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BXN1v67b9f9VHGe3FhaQuD
WHAT: CHANGE.log entries for #167 (Curtis Chambers, no X known) and #165 (Jason Alexander, @jasona verified via x user), roster rows for both, weekly note for Mon 16 Nov. WHY: standing rule, every merged PR gets its change line and roster row the day it merges, with the verified X handle or the honest 'none known'. LR-T: blip, docs, change-log, roster, pr-165, pr-167 LR-D: blip LR-K: change-log-16-nov Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BXN1v68N3DWV2UhtuJggZz
|
Merged in I did the rebase onto #160 in the merge so you did not have to: kept your Still open, and it is not a blocker: no real Android RCS contact has been sent to through the Mac yet. If you get a send to one, please say what chat.db shows (service and error), and I will record it. I deploy this to my own machines today. Generated by Claude Code |
… 14 Nov correction WHAT: WEEKLY-NOTES.md gets its own 2026-W47 section (Mon 16 to Sun 22 Nov). The "Mon 16 Nov: #167 and #165 merged" block moves into it from W46, where it was mis-filed (16 Nov 2026 is a Monday, ISO week 47). W47 also records the audit of all 128 merged PRs: 21 of them (#37 #38 #40 #41 #42 #46 #50 #51 #52 #53 #55 #56 #57 #58 #59 #60 #63 #64 #65 #66 #67, merged 6 to 10 Sep) had no comment and no review from nixfred, so each now carries a specific thank-you on the PR itself (Erik Fillipsveen @Fileri 13, Jon Kinney @jondkinney 7, Danny Cecil @jefehoser 1). W46 gains Jason Alexander's 14 Nov correction to his own #165 evidence, quoted verbatim, which is why #165 stays "not yet proven on a real RCS contact". WHY: standing rule (Fred, 2026-10-31): audit every merged PR at the start of a run and thank any author who got no visible comment; notes are written the day the work lands. VERIFIED: bun test on main, 927 pass, 0 fail, before the comments went out; all 21 comments posted without a failure. OPEN: 2.6.2 (26 Sep) is still the latest release and manifest.json says 2.6.2 while CHANGELOG.md Unreleased holds every PR since; a release is Fred's call. Erik, the largest contributor (22 PRs), is not named in Post 4 or the 2.6.2 release notes. LR-T: blip, weekly-notes, contributors, thank-you, audit, w47 LR-D: blip, docs, github, community LR-K: blip-catchup-thanks-21-prs-2026-11-17, blip-w47-section, jasona-165-correction Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BXN1v3tUKC1p1NZKBgXXvk
…d voice messages WHAT: Unreleased becomes 2.7.0 (2026-11-17) with a short "Before you update" block on top (re-run blip-setup on the Mac, the optional ffmpeg and Qt6 packages for voice and stickers, the opt-in keys replies=on, tapbacks=on and shared_city=on, and the two things not yet proven live). manifest.json and the README version badge go 2.6.2 -> 2.7.0, and the README test badge goes 790 -> 927 passing, the real count. The CHANGELOG heading uses hyphens, not the em-dashes of the older headings, because Fred's no-em-dash rule covers every file; the version-agreement test only needs "## 2.7.0" at the line start. WHY: Fred, 2026-11-17: asked to consider a 2.7.0 release, was shown the readiness check (200 commits from 22 people and 42 merged PRs since 2.6.2, all of it already written up under Unreleased, no open PRs), then ordered the official release. 2.6.2 (26 Sep) was the latest release for seven weeks. VERIFIED: bun test on gus at c95b556, the tree before this commit: 927 pass, 0 fail. HEAD, origin/main and forgejo/main were all c95b556 and gh listed no open PRs. The ui.test.ts version-agreement test passes with the three files in step, and gitleaks found nothing in the diff. The suite on the larry VM shows one failure only because ffmpeg is not installed there (audio-waveform.test.ts shells out to it); gus has ffmpeg and is clean. OPEN: a send to a real RCS contact through the Mac (#165) and a native reply on a 27.2 gateway (#160) are not proven live; both are said so in the notes. CONTEXT: staged by path, the three release files only. TRIAGE-2026-10-02.md and pitch_deck_content.md are untracked and not part of the release. LR-T: blip, release, 2.7.0, changelog, version, contributors LR-D: blip, release, github LR-K: blip-2-7-0-release-2026-11-17, blip-release-after-2-6-2 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BXN1v3LkbABkmYshbwkWb6
…ger say the release is open) WHAT: CHANGE.log gets the 2.7.0 release entry (newest first, feeds X-POSTS.md): tag v2.7.0 on 9fe5d98, the headline PRs, the 21 outside contributors thanked by handle, Erik Fillipsveen named, and the two things not proven live (#165 on a real RCS contact, #160 on a 27.2 gateway). WEEKLY-NOTES.md W47 loses the line "A release is Fred's call", which stopped being true when Fred ordered it, and records the release instead, with a note that the post for it names Erik. WHY: the notes are written the day the work lands, and a notes file that still says the release is open sends the next run off to ask Fred again. VERIFIED: docs only, two files staged by path; gitleaks clean on the diff. The release itself was verified in 9fe5d98: bun test on gus 927 pass 0 fail, origin and forgejo main and the peeled v2.7.0 tag all at 9fe5d98, the GitHub release live and marked Latest, raw manifest.json on main says 2.7.0. OPEN: the 2.7.0 announcement post is not written; Fred posts, Larry drafts. LR-T: blip, release, 2.7.0, change-log, weekly-notes, w47 LR-D: blip, docs, release LR-K: blip-2-7-0-release-2026-11-17, blip-release-after-2-6-2 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BXN1v3LkbABkmYshbwkWb6
What
A 1:1 now sends on the service its own messages say, not on the Mac's chat label. On a macOS 27.2 gateway, every send to two Android contacts who text over RCS failed. One got iMessage, and every send failed with error 22. The other got plain SMS, which the Mac logged as error 4 ("⚠ Not Delivered"). But each of those messages carried the contact's read receipt 1 to 6 seconds later. Both contacts now get RCS. A relayed SMS or RCS in a 1:1 that has a read receipt no longer shows "Not Delivered".
Why
mergeChatslet theservicefromimsg chatsoverride the answersendServiceForMessages()worked out from the window. That field is the chat row'sservice_name, which is a label. On a gateway that keeps one row per conversation, the label follows the Mac's own last send, so trusting it loops.#97 and #147 stopped the label from turning a blue DM green. #151 stopped it from turning a green DM blue when the newest row is our own send. But #151 still let the label move a green DM onto iMessage when the newest row is inbound. That allowance assumed the label is the newest row's service, and it isn't. Here the newest row was the contact's own RCS reply and the label still said iMessage. The other contact's label said SMS, which neither guard covers, so they got plain SMS.
Messages hands SMS to the iPhone to send. For an RCS contact, the iPhone delivers it anyway, but the Mac often records error 4. Those rows had
date_readfrom the recipient's device, and carrier SMS has no read receipts, so they were delivered.Quiet conversations never had evidence in the window, so they always sent on the label.
Changes
mergeChats: a DM with rows under its own id keeps the window's service. The label decides only for the rest. Groups are unchanged, since they send by chat id. This replaces the fix: the chat list never turns a blue DM green #97/Don't let a unified SMS chat row turn an iMessage DM green #147/Don't let the chat list turn a green DM back to iMessage #151 guards.sendServiceForMessages(msgs, prefer, target)decides only from rows whosechatKeyis the id the send goes--to.foldThreadAliaseskeeps the canonical's own service, never an alias's, so a merged phone + email conversation can't hand one handle the other's service.SendState.sendServiceArg.name@rbm.goog) do.thread.ts: reportsservicethroughthreadSendService(), from the opened conversation's rows under its own id. When there are no such rows it reports none, and the label stands. BlipView keeps it with the chat it was loaded for (noteThreadService) and askssendServiceFor()→SendState.sendServiceArgfor text and file sends.SendState.mjsis rebuilt; currentbun buildreproduces main's copy byte for byte.deliveredorread_at) is an iMessage newer than their last text, the DM sends iMessage. An error-22 row is never delivered, so the Don't let the chat list turn a green DM back to iMessage #151 loop stays closed.sendFailed(): a read receipt outranksmessage.error, but only on SMS or RCS rows in a 1:1. An iMessage row or a group row keeps the Mac's verdict. Thread loads are--richand carry the receipt. The collector'srecentrows are not, so its failure toast still fires for a relayed SMS.How it was verified
bun testis green (901 pass). I checked each change by undoing it, and each undo makes a test fail. There are 11 such checks:thread.tsand in the foldSendStateand in the viewscripts/test-send-service-ui.pyis new and added to CI. It runs the shipping BlipView functions under QtTest against the realSendState.mjs. The other Qt UI regressions pass, andqmllintreports nothing new onBlipView.qml.Any send/receive behavior was exercised against my own number only. This PR made no test send. RCS can't be exercised on the self-thread, which is iMessage. These checks against the gateway were read-only, metadata only:
thread.tsreturnsservice: "RCS"for the SMS-labelled contact, which has 0 "Not Delivered" bubbles (6 before).UI change: nothing in the layout changes. The visible difference is a message that was read losing "⚠ Not Delivered". There's no screenshot, because the only examples are real conversations.
Touches an invariant in
CLAUDE.md: adds "A DM sends on what its own messages say, never on the Mac's chat label". In the README'sprefer_imessageparagraph, the failed-iMessage fallback can now be RCS.This overlaps #160 in
thread.tsdecorate(), in BlipView'ssend(), and inSendState.mjs. Whichever lands second needs a small rebase;SendState.mjsregenerates fromsend-state.ts.🤖 Generated with Claude Code