mirror of
https://github.com/logos-co/logos-app.git
synced 2026-08-31 03:11:08 +00:00
This flake now pins nothing by rev. Each pin was checked against the specific
thing its own comment named, not against ancestry — every upstream was
squash-merged, so the old revs are correctly not ancestors of the new masters.
logos-module 9812dc8c -> master. Already EQUAL to master; the pin
was a no-op.
logos-module-loader-qt 06134bf -> acd07cf. Its comment said "drop the rev
once that merges" — logos-module-loader-qt#8 did.
logos-liblogos f2a15ef -> b949212. The comment held it for
liblogos_core's export list, "not on master".
liblogos#177 merged that exact branch
(fix/b4-align-protocol-with-qt-host), and #178
relocked it onto the grant.
logos-capability-module 07dba1f -> c670f7f. Same universal port, now merged
as capability-module#24.
logos-package-manager-ui f135195 -> 6bf1500. package-manager-ui#69 merged;
master's metadata.json reads type: ui_qml,
interface: universal, so the view plugin is there.
VERIFIED BY RUNNING THE GRANT, not by a green build. The universal
capability_module fails CLOSED without its host_services grant, and that failure
is silent: the app launches, the UI renders, and every cross-module call returns
{"result":"","status":"ok"}. checks.host-services-test — which exercises that
path directly — passes, as does checks.smoke-test and `nix build .#default`.
Note for a follow-up: logos-module-loader-qt's comment also says that once #8
landed, "liblogos's own lock can carry it and this input can go away". Only the
pin is retired here; removing the input entirely is a separate change.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
16 MiB
16 MiB
The file is too large to be shown.
View Raw