Repository navigation
feat(gbc): pick a control preset and GBC-Nest from the 解析 panel - #102
Merged
Merged
Conversation
The panel gains 规模 (the shipped small / mid / large control presets) and 程序 (the plain program or GBC-Nest for panels and people). Both are passed as config layers after the generated machine (gbcConfigLayers): a preset that the installed version does not ship is skipped, and a version without configs/programs/gbc_nest.py is refused with a message instead of silently running the plain program. The training batch uses the mid preset and the standard program. Also fixes a real gap: startGbcJob now runs the installed GBC version (settings.gbcVersionId) unless the caller names another, so the panel and the batch no longer fall back to a global gbc on PATH.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What and why
The first slice of P2: the controls the installed KohakuGBC version already ships, exposed in the 解析 panel.
small/mid/large, the shipped control presets, appended as a config layer after the generated machine (gbcConfigLayers).configs/programs/gbc_nest.py), for panels and people. A version without that file is refused with a message rather than silently running the plain program; a preset the version does not ship is skipped.It also fixes a real gap found while wiring this:
startGbcJobignoredsettings.gbcVersionId, so the panel and the training batch could run a globalgbcfromPATHinstead of the installed version. The job now defaults to the installed version and only uses another when the caller names one.How it was verified
pnpm typecheck: all five projects pass.pnpm --filter @clover/desktop exec vitest run --project main test/main/services/gbc.test.ts: 10 passed; the new cases cover an existing preset being added, a missing one skipped, no version meaning no layers, a version without GBC-Nest refused, and the exactrunarguments with the preset then Nest after the machine.pnpm --filter @clover/desktop exec vitest run --project renderer --project shared: 2162 passed, the one failure is the known mock-install environment test.ESLint and Prettier clean;
pnpm buildsucceeds.No e2e change: the mock ignores the two new optional fields and the panel flow is already covered.
pnpm typecheck, lint and Prettier, the suites abovepnpm test:e2e— no new UI path that e2e does not already walkpnpm test:install— not neededDocs updated:
reference/studio-node-board.md§11 (规模 / 程序, the layers, the version default)Platforms affected
Notes for the reviewer
Deferred from P2 with reasons, so the PR stays reviewable:
tag_*controls exist, but we have no installed clone on this machine to confirm where tags land ingraphs.jsonl; guessing a schema we cannot test is worse than leaving it out.gbc hostremote engines: needs the host/SSH management the app does not have for GBC yet..txt(decision in feat(gbc): a 解析 panel in the Studio and a labelling batch in 训练 #99); the graphs are stored per picture, so this is unblocked when the trainer supports it.