122cda84b3
Merge pull request 'Playbooks 07 and 08: read the forge branch's head before trusting an agent's push' (#479) from pb/read-the-pushed-head into main
mesh-tools2026-10-26 10:03:59 +00:00
be1bbe0a65
Playbooks 07 and 08: read the forge head before trusting a reported push, so a try never runs on a push that did not happen
jschoubben2026-10-26 11:02:44 +01:00
585551722a
Merge pull request 'Design 47 follows ADR 0547: node agents compose, the grants part composes who may say what, the bus node writes the user list, and the controller leaves build and delivery in three milestones (issue 1945)' (#478) from fix/1945-user-list-follows-r into main
mesh-tools2026-10-26 07:51:14 +00:00
f1c50d3e01
Design 47: the assigner's acceptance test names its six rules, as ADR 0547 §4 leaves them; the bus node follows the controller's list until the third milestone, for hand-back; placement stays where it is for now (issue 1126)
jschoubben2026-10-26 02:59:58 +01:00
d70791ddee
Design 47: sentences that still named the controller as composer, issuer or reader of grants follow ADR 0547 as amended
jschoubben2026-10-26 02:58:12 +01:00
5bdeb8db83
Design 47: take-over's test as ADR 0547's insight states it
jschoubben2026-10-25 22:21:24 +01:00
d94f8a05a8
Design 47 follows ADR 0547: node agents compose their own modules, a grants part composes who may say what, the bus node writes the user list alone, and the controller leaves build and delivery in three milestones
jschoubben2026-10-25 22:07:56 +01:00
fd0d712f7d
Merge pull request 'Design 47 and the as-is delivery after release 6: acceptance tests on wants, as-is 04 and 13 as the mesh runs now (issue 1130)' (#477) from fix/1130-after-r6 into main
mesh-tools2026-10-26 00:39:21 +00:00
de4c69d476
Name no node in design 47's test 10 proof: hq is public (issue 1130, review round 2)
jschoubben2026-10-26 01:28:32 +01:00
9eb963d5f1
Give each restated acceptance test of design 47 its own proof, and add puts, the hand path and the builds bucket to as-is 04 (issue 1130, review round 1)
jschoubben2026-10-26 01:27:21 +01:00
c255fd071c
Describe delivery after release 6: as-is 04 and 13 follow wants, compositions and haves, design 47's acceptance tests restated (issue 1130)
jschoubben2026-10-26 01:21:42 +01:00
b11120f681
Merge pull request 'Playbooks 07 and 08: lessons from the 2026-10-25 recovery' (#476) from pb/recovery-lessons-2026-10-25 into main
mesh-tools2026-10-25 12:30:43 +00:00
0354947ba6
Playbooks 07 and 08: the dependent-pull-request rule leaves delivery group members out; the instance-search rule is a numbered item
jschoubben2026-10-25 13:29:17 +01:00
4d596bd211
Playbooks 07 and 08: open a dependent pull request only after delivery, fold a follow-up into the unapproved one, search every instance of a fault kind
jschoubben2026-10-25 13:27:54 +01:00
0e1d0392bc
Merge pull request 'R6: design 47, the glossary and playbook 07 follow the switch to wants (issue 1130)' (#475) from feat/r6 into main
mesh-tools2026-10-24 23:00:22 +00:00
ba09497092
Name the hand-path command mesh-hand-compose, as mesh-sdk feat/r6 now calls it: main's mesh-compose is the composer (ADR 0482, issue 1943)
jschoubben2026-10-24 23:06:00 +02:00
2d3a205740
Merge remote-tracking branch 'origin/main' into feat/r6
jschoubben2026-10-24 23:00:09 +02:00
49dd8fc7e3
Merge pull request 'R4g finish: hq's checks as Go tests, the old word struck, the marker (issues 1296, 1935)' (#473) from feat/r4g-finish into main
jschoubben2026-10-24 20:56:42 +00:00
9558b71a4a
Merge pull request 'Playbook 07: read a verdict's trunk against main at once, because a lagging source revision makes the Approve merge nothing (issue 1726)' (#474) from pb/verdict-trunk into main
jschoubben2026-10-24 19:47:56 +00:00
3dbb5d4e65
Design 47: the switch needs release 5's bus grants and the operator's key first, the record hold's exception after the switch, and nothing checks the budgets (review of 4b514e4, issue 1130)
jschoubben2026-10-24 21:06:41 +02:00
4b514e4d69
Design 47, the glossary and playbooks 04, 07 and 08 for release 6: wants and signed compositions, no plans (ADR 0539, issue 1130)
jschoubben2026-10-24 21:01:24 +02:00
8433e42c03
Playbook 07: the glossary's words, the source revision and the gate (review)
jschoubben2026-10-24 13:54:31 +02:00
129af9906d
Playbook 07: read a verdict's trunk against main at once, because a lagging trunk makes the Approve merge nothing (issue 1726)
jschoubben2026-10-24 13:50:37 +02:00
c12d503b2e
Merge pull request 'Playbooks 07 and 08: ask who ends each bus key, how a switch-over goes one node first, how a group's integration level runs; a hand merge runs it first (research 071, issue 1941)' (#472) from pb/research-071-checks into main
jschoubben2026-10-24 11:10:51 +00:00
8498006d23
Playbooks 07 and 08: answer the review (no lab, plain words, checks on the forge, issue 1941)
jschoubben2026-10-24 12:50:35 +02:00
d20ed892cf
Playbooks 07 and 08: three questions in the milestone file, and a try before a merge by hand (research 071, issue 1921)
jschoubben2026-10-24 12:45:57 +02:00
d08cac4e86
Answer the review of hq's finish: role names in the as-is delivery, ADR 0524-0529 named, stale check names fixed (issue 1296)
jschoubben2026-10-24 04:40:02 +02:00
ca460b681b
hq's checks read the whole repository: mark the package reads-outside so its tests always run (issue 1935)
jschoubben2026-10-24 03:46:56 +02:00
b3b778a1a6
Describe delivery as it runs after R4g: the parts, the build modules, the one check (issue 1296, H9)
jschoubben2026-10-24 03:41:01 +02:00
6dc008b5e7
Delete merge-check.sh and the four Python checks; the Go tests replace them (issue 1296, H6)
jschoubben2026-10-24 03:39:11 +02:00
82aa3f6afa
Merge pull request 'Playbooks 07 and 08: after a failed try, a local run of the integration level is a good first step (guideline)' (#471) from pb/try-fails-run-locally into main
jschoubben2026-10-23 21:50:24 +00:00
8055006ba1
Playbook 07: the two 'never run integration tests on the machine' lines say 'while developing', so the guideline after a failed try no longer contradicts them (review-471)
jschoubben2026-10-23 23:24:15 +02:00
2b53b6a5d2
Playbooks 07 and 08: say the local run after a failed try as a guideline, not a rule (the operator)
jschoubben2026-10-23 23:22:21 +02:00
87c338da8e
Playbooks 07 and 08: a failed try or check is run locally at once, so every failure shows in one run, not one per try (the operator, 2026-10-23)
jschoubben2026-10-23 23:20:27 +02:00
656f3af86f
Merge pull request 'R4g+R5 (hq): its checks as Go tests; design 47, the glossary and playbooks say what R4g builds (design 47)' (#470) from feat/r4g into main
jschoubben2026-10-23 17:54:31 +00:00
e8bd9171cd
Design 47: the gate reads mesh/check, the forge requires no status (ADR 0534 §1 changes ADR 0447 §5)
jschoubben2026-10-23 16:31:15 +02:00
8f7c23da7d
Make a timer that writes state, or a rule in code that overrides a person, a change the review asks (issue 1903)
jschoubben2026-10-23 16:10:44 +02:00
d49dccb8d5
Use the glossary's word composition check where the merge gate was struck (issue 1303)
jschoubben2026-10-23 16:10:08 +02:00
6ead37ee7b
Design 47: no status is required on any repository, so hq's Python checks go after the delivery (ADR 0534), and ADR 0529 is named
jschoubben2026-10-23 16:10:01 +02:00
c8ccbfe35f
Playbooks 07, 08 and the overview name what runs: the checker's try and status, hq's Go tests, the stream backlog, and the way to deliver without the builder (issue 1888)
jschoubben2026-10-23 15:17:30 +02:00
99fa8f0f09
Glossary: the build seat, the build queue and the build capacity as asks on a language's streams, and the build module (issue 1788)
jschoubben2026-10-23 15:16:08 +02:00
75056fa931
Design 47 says what R4g and R5 build as one milestone: asks on streams, the assigner's wants, hq's Go tests, the hand path (issue 1130)
jschoubben2026-10-23 15:15:18 +02:00
47e78d8ff4
Strike the word merge gate: the composition check replaces it in the glossary, the designs and the playbook, and old issues keep it dated (issue 1303)
jschoubben2026-10-23 15:12:57 +02:00
1385452dc9
Say in the checks' README that the same four checks are Go tests (issue 1296)
jschoubben2026-10-23 15:11:20 +02:00
5b63a86464
Run words.py as Go tests, with the clock injected and the catalogue comparison skipped when unset (issue 1296)
jschoubben2026-10-23 15:10:53 +02:00
344c92498d
Run cycle.py as Go tests: a design's layer, status, decisions and code, and the issues folder's numbers and frontmatter (issue 1296)
jschoubben2026-10-23 15:07:48 +02:00
9339e953b8
Run index.py as Go tests: the reading order's topics are written and no list of records is stored (issue 1296)
jschoubben2026-10-23 15:07:48 +02:00
0ad95768b6
Run records.py's five checks as Go tests, each proven against a broken tree first (issue 1296)
jschoubben2026-10-23 15:07:07 +02:00
dd45f31307
Give hq a Go module and the checks' shared reading, so the four checks can become Go tests (issue 1296)
jschoubben2026-10-23 15:06:25 +02:00
fa7629d02b
Merge pull request 'Design 42: the package manager and kernel rows as ADR 0507, 0508 and 0510 decide (upgrade safety)' (#467) from feat/upgrade-safety into main
mesh-tools2026-10-22 18:34:11 +00:00
93963aeb51
Merge pull request 'Playbooks: how a module's old directory is closed, and a refused try (issues 1843, 1854)' (#468) from pb/moving-a-module-directory into main
mesh-tools2026-10-22 18:19:19 +00:00
795b3b5d45
Merge pull request 'Design 63 (the mail web UI with mail reading and Owly) and playbook 10 (visual design)' (#469) from design-63-mail-web-ui into main
mesh-tools2026-10-22 18:16:54 +00:00
030e985694
Answer the review's second round: the mockup's sheet holds invented sales figures; milestone 5's row lists only teaching and the counts
jochen2026-10-22 19:21:38 +02:00
eddd7ba684
Answer the review: mockup data invented throughout, later milestones' scope marked proposed, design 59 says where it is replaced, playbook 10 aligned with 04 and 07
jochen2026-10-22 19:19:52 +02:00
5d745d04db
Write design 63 and playbook 10, so the mail page's next steps are built from a picked design
jochen2026-10-22 18:42:42 +02:00
7fcca88c01
Playbooks: how a module's old directory is closed, and a refused try (issues 1843, 1854)
jochen2026-10-22 18:25:05 +02:00
333e73830c
Amend design 42 for the package view, one-at-a-time upgrades and the fallback kernel (ADR 0507, ADR 0508, ADR 0510; issues 1840, 1841)
jschoubben2026-10-22 17:41:15 +02:00
58555ec330
Merge pull request 'Name the session that pushed a head in its pull request (ADR 0505 §5)' (#466) from feat/quiet-deliveries into main
mesh-tools2026-10-22 14:01:16 +00:00
9a74de0bd6
Name the session that pushed a head in its pull request, so a head that needs a new one names its author (ADR 0505 §5)
jschoubben2026-10-22 14:25:50 +02:00
77d4a8a39e
Merge pull request 'Never cut a milestone by an order between its parts; the delivery plan keeps that order' (#465) from process/07-order-is-the-plans into main
mesh-tools2026-10-21 23:44:17 +00:00
2b857df4b5
Never cut a milestone by an order between its parts: the delivery plan keeps that order from the module graph
jschoubben2026-10-22 01:30:20 +02:00
73a06324dc
Merge pull request 'Correct design 47 after the review of #463' (#464) from design/47-cites-fixes into main
mesh-tools2026-10-21 21:15:04 +00:00
c07483081d
Cite ADR 0494 in design 47: the checker's milestone is one group, not two deliveries
jschoubben2026-10-21 22:54:18 +02:00
1955d7f624
Correct design 47 after review of #463: R4e timing of the SDK group, the wait list's Waited-by condition, and what builds images from R4g
jschoubben2026-10-21 22:52:18 +02:00
57396112c7
Merge pull request 'Name the accepted decisions 0480 to 0493 in design 47 and state what they changed' (#463) from design/47-cites-0484-0493 into main
mesh-tools2026-10-21 20:49:11 +00:00
38020bee9a
Name the accepted decisions 0480 to 0493 in design 47 and state in its prose what they changed, so the design matches what holds
jschoubben2026-10-21 22:44:29 +02:00
04fff52d3b
Merge pull request 'Design 36: the plugin's session-start and Stop hooks beside the guard, and the waits mirror in the agent's home (ADR 0489, ADR 0490)' (#462) from design/36-wait-list-hooks into main
mesh-tools2026-10-21 17:53:42 +00:00
ac72fe6422
Design 36: the waits mirror's file is named by Claude Code's session id
jochen2026-10-21 19:30:21 +02:00
9483caae2d
Design 36: the waits mirror is the module's own entry in the agent's home; the hook tools refuse mesh and node scope
jochen2026-10-21 19:29:38 +02:00
302ab2856a
Design 36: the plugin's session-start and Stop hooks beside the guard (ADR 0489, ADR 0490)
jochen2026-10-21 19:27:33 +02:00
a214776696
Merge pull request 'Playbook 07: a new verb of a seat the mesh defines waits for the SDK and a controller delivered with it' (#460) from playbook/seat-contract-order into main
mesh-tools2026-10-21 16:14:23 +00:00
ac7e445e7a
Merge pull request 'Playbooks 07 and 08: a milestone is one delivery group released by one plan; a fix found during it rides in the group' (#461) from playbook/08-release-per-milestone into main
mesh-tools2026-10-21 16:13:04 +00:00
0162e04f92
Describe a milestone as one delivery group released by one plan, and make a fix found during it ride in the group
jschoubben2026-10-21 18:08:58 +02:00
37a01b93a5
Use the glossary's word for the milestone's branch
jschoubben2026-10-21 18:05:07 +02:00
c6fc663019
Make a small fix in mesh-sdk or mesh-controller ride with the milestone's release, so each is delivered once per milestone
jschoubben2026-10-21 18:04:49 +02:00
5f281dd2a3
Playbook 07: a new verb of a seat the mesh defines waits for the SDK and a controller delivered with it, so the prepare step searches the SDK for each seat a change adds verbs to
jschoubben2026-10-21 15:22:27 +02:00
d5d8212204
Merge pull request 'Playbooks 07 and 08: the session that opens an issue along the way fixes it' (#459) from playbook/fix-what-you-open into main
mesh-tools2026-10-21 12:47:52 +00:00
aa56e9b9e7
Playbook 07: the fixed scope and fixing what you open agree, a blocking issue still goes in
jochen2026-10-21 14:42:27 +02:00
4f8c469329
Playbooks 07 and 08: the session that opens an issue along the way fixes it
jochen2026-10-21 14:40:56 +02:00
abdd364f09
Merge pull request 'Playbook 07: name a pull request's own issues as 'issue <n>', and no other' (#458) from playbook/pr-issue-words into main
mesh-tools2026-10-21 12:17:55 +00:00
ad0a95dc0c
Playbook 07: the plural 'issues <n>' is read too; 1654 was another milestone's issue
jochen2026-10-21 14:06:36 +02:00
c770bc1c50
Playbook 07: name a pull request's own issues as 'issue <n>', and no other
jochen2026-10-21 14:05:28 +02:00
ab3ac686dc
Merge pull request 'Correct design 47's R4c row (installed toolchain) and move 1305 from R4e to R4f' (#457) from design/47-r4c-row into main
mesh-tools2026-10-21 11:40:39 +00:00
373b7e7dd1
Correct design 47's R4c row (installed toolchain, ADR 0477/0478) and move 1305 from R4e to R4f
jschoubben2026-10-21 13:35:58 +02:00
6eda564701
Merge pull request 'Add milestone F2b to design 60, and make every strict reader of manifests a release order in playbook 07' (#456) from design/60-milestone-f2b into main
mesh-tools2026-10-21 11:00:57 +00:00
7c8e1f521d
Add milestone F2b to design 60, and write into playbook 07's prepare step that every strict reader of manifests ships a new word first
jschoubben2026-10-21 12:53:37 +02:00
a1188b64a4
Merge pull request 'Mark design 61 designed now that ADR 0479 is accepted, and follow its last version' (#455) from design/61-adr-0479-accepted into main
mesh-tools2026-10-21 10:40:20 +00:00
6db031f486
Merge pull request 'Design 62: name ADR 0481 accepted, and say the page reaches OkiOki only over the bus' (#454) from design/62-review-fixes into main
mesh-tools2026-10-21 10:39:21 +00:00
be30b63976
Mark design 61 designed now that ADR 0479 is accepted, and follow its last version: PDF and JPEG only, the printer asks its own link
jschoubben2026-10-21 12:36:24 +02:00
993c7def3a
Design 62: name ADR 0481 accepted, and say the page reaches OkiOki only over the bus
jochen2026-10-21 12:32:31 +02:00
a2918df0aa
Merge pull request 'Design 62: the bookkeeping portal (ADR 0481)' (#453) from design/62-bookkeeping-portal into main
mesh-tools2026-10-21 10:31:02 +00:00
46d6840c67
Design 62: the page is read-only and its reach internal before assignment (ADR 0481 v4)
jochen2026-10-21 11:56:03 +02:00
82343be78f
Merge pull request 'Write design 61, the print and scan page, as the home of ADR 0479' (#452) from design/61-print-and-scan into main
mesh-tools2026-10-21 09:32:15 +00:00
75f2cbb6d6
Merge pull request 'Name the accepted ADR 0468 in design 60 and break its building into three milestones' (#451) from design/60-adr-0468 into main
mesh-tools2026-10-21 09:27:16 +00:00
467a2e6c97
Credit restart and the settings read to ADR 0467, and only configure to ADR 0474
jschoubben2026-10-21 11:18:56 +02:00
115df9fd69
Follow ADR 0479 version 6: the glass only until the feeder is decided, and how the upload bound holds
jschoubben2026-10-21 11:18:13 +02:00
c7bebf53d4
Answer the review: the page follows ADR 0474, and records who managed the printer
jschoubben2026-10-21 11:16:48 +02:00
e0118ca67e
Write design 61, the print and scan page, as the home of ADR 0479
jschoubben2026-10-21 11:15:36 +02:00