Skip to content

Commit b63dc4e

Browse files
committed
docs(design): the four upstream asks, landed in every repository, and what the sandbox measured
1 parent 04a64bf commit b63dc4e

1 file changed

Lines changed: 38 additions & 1 deletion

File tree

.agents/docs/2026-09-13-four-upstream-asks-from-a-ui-framework.md

Lines changed: 38 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -5,7 +5,7 @@ status: landed
55

66
# Four upstream asks from a UI framework: what each one is under mcpp's design, and the combined plan
77

8-
**Status:** the mcpp half landed in 2026.9.13.1 (PR #629, released and indexed 2026-09-13); the index half landed (openxlings/xim-pkgindex#832); the mcpp-plugins half is in review as 0.9.0 (mcpp-plugins#20). §8.1 to §8.3 record what implementation changed.
8+
**Status:** landed on 2026-09-13 in every repository. mcpp 2026.9.13.1 (PR #629, released, mirrored on both ends byte-identically, indexed by openxlings/xim-pkgindex#833); the Cubism recipes (openxlings/xim-pkgindex#832); mcpp-plugins 0.9.0 (mcpp-plugins#20, tag `v0.9.0`, GitCode mirror byte-identical, indexed by mcpplibs/mcpp-index#405). §8.1 to §8.4 record what implementation changed and what the sandbox measured.
99
The first ask was analysed in an earlier draft of this record (the action
1010
buffer, §2); this revision reads all four against the rules the 2026-09-11
1111
and 2026-09-12 records state, and replaces that draft. The review accepted
@@ -579,6 +579,43 @@ Four additions came from the review and are folded in above:
579579
and is filed there (mcpplibs/openkal-windows#19) rather than worked around
580580
in this workflow.
581581

582+
### 8.4 What the sandbox measured (2026-09-13)
583+
584+
`xlings subos use eco-2026-9-13-1 --sandbox`, both mirrors set to CN, the
585+
engine addressed by its store path and pinned, every package from the
586+
published indexes and nothing from a working tree. Thirteen assertions, none
587+
failed, none skipped:
588+
589+
- `xim:mcpp@2026.9.13.1` installs from the index and reports its version at
590+
the store path; the mcpp-side mirror reads back as CN.
591+
- An action with 600 inputs and 600 outputs: `build.ninja`'s edge carries
592+
all 600 outputs and the program runs.
593+
- `${mcpp.self} stage` copies the linked program byte for byte into a
594+
directory the action never created; the token does not reach
595+
`build.ninja` literally.
596+
- `mcpp:plugins@0.9.0`, `dist-web`: `web/` carries `index.html`, the `.js`
597+
launcher and the `.wasm`; the copy edges name the engine's `stage`; a
598+
second pack rewrites nothing.
599+
- `mcpp:plugins@0.9.0`, `dist-apk`: a project template's permission,
600+
receiver and application id reach the manifest; a signed `.apk` is
601+
produced.
602+
- `cubism-sdk-native` and `cubism-sdk-web` install from the anonymous
603+
upstream URL; `Core/LICENSE.md` is present in both.
604+
605+
The first run of the same script, before 0.9.0 was indexed, failed the two
606+
plugin sections with `E_NOT_FOUND: package 'mcpp:plugins@0.9.0'` and passed
607+
the other eight, which is the negative direction for the index step.
608+
609+
What the release chain measured on the way: the four platform archives are
610+
byte-identical on the upstream release, the GitHub mirror and the GitCode
611+
mirror (GET, then `sha256`), and the four hashes in the index bump PR equal
612+
those; the plugins source archive on GitCode is `cmp`-equal to GitHub's tag
613+
archive. Main's CI on the bootstrap-pin commit first failed in every job
614+
that bootstraps, with `package 'mcpp@2026.9.13.1' not found in the synced
615+
index (xim@artifact:dafbe58)`: the runners read the previous index
616+
artifact although both pointer files already named `9246b34`, and reruns
617+
after the propagation window passed.
618+
582619
One observation stands as recorded and not acted on: `options::resources`
583620
as a single directory declared as an input (§3.4). It affects HuxerUI
584621
little today, since the framework's Android library carries no `res/` and

0 commit comments

Comments
 (0)