mcptask-rails-runner 0.3.20 → 0.3.22
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/CHANGELOG.md +99 -1
- data/lib/mcptask_rails_runner/version.rb +1 -1
- metadata +1 -1
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA256:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: 17952c83fadf385cdd9425762619050d2a287af1ddea4f678ea75f6ebfd1d385
|
|
4
|
+
data.tar.gz: fa664976f9e79876db7e87083f8cedff01f42cef9e2c161bb997fd8fa459a47b
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: 503bceb521790003682c224959f958aab8eae986c8207f31b90636eda9353f4be3e951161d22e4c16a6d2fe9aa0963506d5621879bdaec61160bb59f5f4e1117
|
|
7
|
+
data.tar.gz: 235729f6fe98e5b68daa6b122deb89077baf085301cf7e6be23284bbff5ad230e01bfb81d30c623fb614b77e4390db01a2cf10521936d6d9b62d44cfdb43bd94
|
data/CHANGELOG.md
CHANGED
|
@@ -5,7 +5,105 @@ entries below describe wrapper changes only. Binary changes are listed in the
|
|
|
5
5
|
[mcptask-releases](https://github.com/jchsoft/mcptask-releases/releases)
|
|
6
6
|
release notes for the same tag.
|
|
7
7
|
|
|
8
|
-
##
|
|
8
|
+
## 0.3.22
|
|
9
|
+
|
|
10
|
+
No wrapper changes. Full notes in
|
|
11
|
+
[mcptask-releases v0.3.22](https://github.com/jchsoft/mcptask-releases/releases/tag/v0.3.22).
|
|
12
|
+
|
|
13
|
+
**Upgrading takes two commands this time, not one.** The usual
|
|
14
|
+
`mcptask_runner update --self` replaces the binary, and that is enough for the
|
|
15
|
+
loop changes below. It is *not* enough for the skills and helpers: those are a
|
|
16
|
+
data pack the binary carries, and they are only written into a project by
|
|
17
|
+
|
|
18
|
+
```
|
|
19
|
+
mcptask_runner update
|
|
20
|
+
```
|
|
21
|
+
|
|
22
|
+
Run it in each host project after updating the binary. Skipping it leaves the
|
|
23
|
+
old `/test-runner` and `/ci-runner` in place, and on a project that is not
|
|
24
|
+
Rails those name commands it cannot run — which is the very thing this version
|
|
25
|
+
fixes. A host that has locally edited a skill keeps its edit; `FORCE=1` takes
|
|
26
|
+
the new one and leaves a `.bak`.
|
|
27
|
+
|
|
28
|
+
Binary changes carried by this version:
|
|
29
|
+
|
|
30
|
+
- **The bundled CI and test skills stop naming a framework.** `/test-runner`
|
|
31
|
+
and `/ci-runner` used to spell out `bin/rails test`, `test:system` and
|
|
32
|
+
RuboCop, and `check_test_lock` recognised a running suite by six regexes,
|
|
33
|
+
five of which were Ruby. On a Django, Node or Swift host the commands did
|
|
34
|
+
not run and the concurrency guard waved every real test command through —
|
|
35
|
+
blind on exactly the projects a universal runner exists for. Both now read
|
|
36
|
+
what the project declares in `.claude/test-commands.json`, which
|
|
37
|
+
`mcptask_runner init` writes and `update` backfills; when the file is absent
|
|
38
|
+
the toolchain is detected from `go.mod`, `Gemfile`, `package.json`,
|
|
39
|
+
`manage.py`, `pyproject.toml` or `Package.swift`, and when nothing is
|
|
40
|
+
recognised the runner says so instead of guessing a command.
|
|
41
|
+
|
|
42
|
+
- **Two more places the prompt spoke Rails to everyone.** The auto-squash
|
|
43
|
+
workflow told every project to run `bin/rails assets:precompile
|
|
44
|
+
RAILS_ENV=test` between its unit and system suites, and the manual
|
|
45
|
+
screenshot step named a Rails class. Neither is measured by what it breaks —
|
|
46
|
+
the first costs a turn or invites a plausible substitute, the second is
|
|
47
|
+
silently ignored — but both contradict the claim the product makes.
|
|
48
|
+
|
|
49
|
+
- **A failed task now waits an hour, not the rest of the day.** A task that
|
|
50
|
+
fails is set aside and the runner moves on; the day no longer ends over one
|
|
51
|
+
piece.
|
|
52
|
+
|
|
53
|
+
- **The dashboard can take a piece off the skip list.** A piece set aside
|
|
54
|
+
earlier in the day can be released from the card, and the runner picks it up
|
|
55
|
+
again without a restart.
|
|
56
|
+
|
|
57
|
+
- **A closed socket is described by whoever actually saw it close.** Two paths
|
|
58
|
+
reported a dropped dashboard connection — the read loop, which knows the
|
|
59
|
+
close code the server sent, and a failed write, which knows only that the
|
|
60
|
+
socket was gone. Whichever finished first wrote the notice, so a close with
|
|
61
|
+
a code could be reported as a bare transport error. The reader's account now
|
|
62
|
+
wins, which is what makes "the server closed it, code 1001" trustworthy when
|
|
63
|
+
you are deciding whether to go and look at the server.
|
|
64
|
+
|
|
65
|
+
## 0.3.21
|
|
66
|
+
|
|
67
|
+
No wrapper changes. Full notes in
|
|
68
|
+
[mcptask-releases v0.3.21](https://github.com/jchsoft/mcptask-releases/releases/tag/v0.3.21).
|
|
69
|
+
|
|
70
|
+
**Upgrading:** run `mcptask_runner update --self` on each host. Both changes are
|
|
71
|
+
binary behaviour, and the first of them only stops costing you a working day
|
|
72
|
+
once the host is actually running the new binary.
|
|
73
|
+
|
|
74
|
+
Binary changes carried by this version:
|
|
75
|
+
|
|
76
|
+
- A deploy of mcptask.online now costs one attempt instead of the working day.
|
|
77
|
+
While the site is being deployed the MCP server is unreachable for minutes and
|
|
78
|
+
then comes back by itself. The runner used to spend its entire recovery budget
|
|
79
|
+
inside about forty seconds — a child dies in ten, five seconds between
|
|
80
|
+
restarts, two restarts — so all three attempts lost the same race before the
|
|
81
|
+
server could possibly be back, and the run ended with `error`. That status is
|
|
82
|
+
what ends the day, so one deploy could end a day that had already completed
|
|
83
|
+
eighteen tasks.
|
|
84
|
+
|
|
85
|
+
The two situations that reach that path are now told apart where they are
|
|
86
|
+
detected, rather than by reading the message afterwards. A deferred tool that
|
|
87
|
+
never loaded is the child's own miss: it keeps its two fast restarts and still
|
|
88
|
+
ends in an error, because it is not going to fix itself. An outage gets six
|
|
89
|
+
restarts at twelve times the wait — about six minutes, which spans a deploy —
|
|
90
|
+
and when that runs out it ends the way a stall does: a verdict the loop
|
|
91
|
+
carries on from, the task left `in_progress` for the next triage, and no bug
|
|
92
|
+
piece, because a deploy is not a defect anybody can fix. The wait keeps the
|
|
93
|
+
heartbeat going, so a card watched through a deploy says the runner is waiting
|
|
94
|
+
rather than falling silent.
|
|
95
|
+
|
|
96
|
+
Like every binary fix, it reaches a host only once that host runs the new
|
|
97
|
+
binary — `mcptask_runner update --self` — and until then a deploy goes on
|
|
98
|
+
costing the day.
|
|
99
|
+
|
|
100
|
+
- Today's skip list is visible. A task the runner has set aside — out of scope,
|
|
101
|
+
failed, already taken by somebody else, a merge it cannot finish — is skipped
|
|
102
|
+
by every later triage that day, and until now the only trace of that decision
|
|
103
|
+
was one line in the log at the moment it was made. It now shows on the startup
|
|
104
|
+
banner, in every wait, and on the dashboard card, so an operator looking at a
|
|
105
|
+
quiet runner can see what it has already declined rather than wondering why a
|
|
106
|
+
piece keeps not being picked up.
|
|
9
107
|
|
|
10
108
|
## 0.3.20
|
|
11
109
|
|
|
@@ -10,5 +10,5 @@ module McptaskRailsRunner
|
|
|
10
10
|
# Binary::INSTALL_DIR. So the pin is per project and per foreground rake
|
|
11
11
|
# invocation; the installed copy is per machine, and the newest binary any of
|
|
12
12
|
# those projects offers is the one that stays.
|
|
13
|
-
VERSION = "0.3.
|
|
13
|
+
VERSION = "0.3.22"
|
|
14
14
|
end
|