Repository navigation
fix: a Luban system's root is declared and its /xlings upstream is its store (2026.10.22.1) - #831
Merged
Merged
Conversation
…s store (2026.10.22.1) In a Luban system the compiler's sysroot / is the system itself (os-release ID=luban, /usr/lib/xlings-abi), and [xlings] upstream = ["/xlings"] reads the home its root record (/etc/xlings/usr) names when /xlings is not there: a system made on a host keeps its home at the path of the home that made it. Verified in a Luban KVM machine: mcpp self doctor without the sysroot or upstream warning, a first mcpp run of a new project in 4 s. Co-authored-by: speak-agent <248744407+speak-agent@users.noreply.github.com>
… (declared_system_root, system_home) -- mcpp names none of xlings' paths Co-authored-by: speak-agent <248744407+speak-agent@users.noreply.github.com>
Sunrisepeak
pushed a commit
to openxlings/xlings
that referenced
this pull request
Oct 22, 2026
…er virtio-fs, the bus, observability, xdesk, luban-os-ci; 2026.10.22.1 (#693) Luban OS ecosystem usability (design `.agents/docs/2026-10-21-luban-os-ecosystem-usability-design.md`), xlings 2026.10.22.1. One commit per checkpoint; with xim-pkgindex#1022, mcpp-community/mcpp#831 and luban-os/luban-os-ci#1. **xlings in a system** - K1: recipes are read in a Luban machine (the mount table names the filesystem; Landlock on uncached 9p; worker scratch local); an unset `XDG_DATA_DIRS` keeps its default. - K2/K4: one place decides where a change goes (`xlings.subos.situation`); a machine's broker carries `sudo xlings install/remove/update` and a user's store addition to the host; a user's own versions need no sudo, `xlings update` says `sudo xlings update`. - A user's view starts with the system's subos description (its doctor is clean); what xlings sets for a session (prompt facts) passes whatever the policy; an `/etc` file deleted under its mount is left out of a sandbox instead of failing it. **Machines** - K6: files over virtio-fs (a virtiofsd a share, `xim:virtiofsd` installed the first time, `--sandbox namespace` then `none`), 9p where it cannot be (no KVM, no daemon); `luban config --files`, said in `luban status`. Measured: `find /usr` 3955 → 359 ms, reading libraries 1041 → 130 ms. - Commands at once each run to their end; a machine started for commands powers off when the last ends (leases, under one lock); a command whose machine goes away says so; `--user` with a name the system lacks is refused, never root. - An archive's packages come along again (import read `{active, installed}` entries as versions and skipped all; the export now names each program's package); a sudo install survives `luban upgrade`. - K14: WSL's `/dev/dxg` probed and said. **The bus, observability, desktop** - openbus/XBus: routers, nodes, links across machines (`luban status --bus`), streams and regions, telemetry on every node (§8.3); a frame begun is finished before any other (macOS's small socket buffers). - The telemetry kit (metrics, flight recorder, logs, spans, Chrome traces); the i18n kit shared by xlings and xdesk. - K7: frames compressed across a narrow channel (opensal 0.6), damage merged. - K11/K13: xdesk bookmarks, collapse, minimap, title bars, free-window resize; the tour; every layout, window and interaction drivable headless and covered (`test_xdesk_interaction`). **CI and checks** - P2's xlings side: package checks `run`, `launcher`, `gui`, `sound`. - K15: the release candidate is dispatched to luban-os-ci's L2 when `LUBAN_OS_CI_TOKEN` is there; luban-os-ci's first local runs found the bugs above. - Recorded for later (design §15): K5 (`--store own`), J32 (a machine's new users cannot own their homes), J33 (a user view's install is active for everyone). Co-authored-by: speak-agent <248744407+speak-agent@users.noreply.github.com>
Member
Author
|
The one red check, Co-authored-by: speak-agent 248744407+speak-agent@users.noreply.github.com |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
In a Luban system (xlings design
.agents/docs/2026-10-21-luban-os-ecosystem-usability-design.md§3.7, J5, M1):/is the system's own declared root (ID=lubanin/etc/os-releaseand/usr/lib/xlings-abi), somcpp self doctorno longer says it belongs to neither the mcpp home nor the project;[xlings] upstream = ["/xlings"](the editions'/etc/mcpp/config.toml) reads the store the system's root record (/etc/xlings/usr) names when/xlingsis not there -- a system made on a host keeps its home at the path of the home that made it.New module
mcpp.luban(a leaf over std) holds both facts;tests/unit/test_luban_root.cppcovers them against fixture trees.Verified in a Luban KVM machine (Luban Desktop):
mcpp self doctorshows neither warning, and a firstmcpp new demo && mcpp runtakes 4 s, reusing the system's gcc.Co-authored-by: speak-agent 248744407+speak-agent@users.noreply.github.com