chore[notask]: align speech, translation and diffusion addons - #3900
chore[notask]: align speech, translation and diffusion addons#3900donriddo wants to merge 1 commit into
Conversation
License compliance — findings detected (warn-only)Critical: 0 · High: 4 · Medium: 0
How to resolve a blocking finding:
Warn-only (shadow) mode — this check does not block merges yet. Updated automatically by the canonical license compliance workflow. NOTICE presence (advisory)Missing NOTICE (advisory, does not block):
|
Review StatusCurrent Status: ❌ PENDING Pending reviews: Needs 1 Management or Team Lead, and 1 more from Management, Team Lead, or Member. |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
Raise the remaining bundled addons to their fabric-aligned releases so the native bundle links one backend set: bci-whispercpp 0.7.1, audiogen-ggml 0.2.2, diffusion-cpp 0.20.0, and tts-ggml floored at 0.7.2 (the registry-client 0.6.1 line that resolves the hyperdb dedup). The llama-family, translation-nmtcpp, and the tts pace / contract / Python-client changes are already on main.
QVAC E2E —
|
QVAC E2E —
|
|
Closing as redundant: the addon alignment bumps have been folded into #3682 (the multi-job PR) per the team decision, so #3682 now carries the multi-job feature plus the full aligned bundle (bci 0.7.1, audiogen 0.2.2, diffusion 0.20.0, tts 0.7.2, and the llama-family/tts-pace from main). No need for a separate alignment PR. |
🎯 What problem does this PR solve?
#3902 aligned the llama-family addons on the fabric-10069 backend set and #3857 brought
tts-ggmlup. The remaining bundled addons (bci-whispercpp,audiogen-ggml,diffusion-cpp) are still on older fabric / addon-cpp builds, so on mobile the flattened native lib dir keeps a mixed backend set — the iOS JsAsyncTask crash blocking the integration PRs (e.g. #3682): the app loads the LLM, then dies loading the embedding model.📝 How does it solve it?
Bumps the still-behind addons to their fabric-aligned releases:
@qvac/bci-whispercpp@qvac/audiogen-ggml@qvac/diffusion-cppIt also floors
@qvac/tts-ggmlat 0.7.2 (was^0.7.0), theregistry-client@^0.6.1line, so the hyperdb dedup fix is guaranteed rather than resolved by chance.Each was checked against actual SDK usage, not just types: none introduce an SDK-facing API break. The tts
pacenarrowing already landed on main via #3857, so this PR is purely the version bumps.🧪 How was it tested?
Typecheck clean;
test-e2e-fullrun for the mobile lanes to confirm the app survives the LLM → embedding transition that crashes today.