Feat/generic diffusers workbench - #5
Merged
Merged
Conversation
Publish the validated MoDiff-client source at df50a30976490c719ac6dc657cacdae28511e651 and mirror the workbench plan status. Expert starts in the generic Stages catalog, Auto hides implementation-only HF entries, and Advanced retains full implementation discovery. The bundle is generated from client dist, with backend-owned custom fields and installed Gallery assets preserved. No backend executor or model-cache behavior changes in this increment. Validation: - Client npm ci, npm run check, npm run check:ui (148 browser tests) - uvx --from ruff==0.12.7 ruff check . --select E9,F - uv pip check --python .venv/bin/python - ./scripts/with-runtime-env.sh ./.venv/bin/python -m modiff.preflight --json --check-port 8088 --fail-on-error - ./scripts/with-runtime-env.sh ./.venv/bin/python -m pytest -q (3038 passed, 509 skipped; 7300 subtests passed) - Fresh WebServer with isolated data: HTTP 200 and exact dist bytes for /, /assets/index.js, and /assets/NodeList.js - Checksum-based dist/web comparison, excluding user and template-gallery Skipped tests and HTTP smoke do not establish live model-generation proof.
Mirror the validated MoDiff-client build from 7320ab5. Preserve backend-owned user fields and local Gallery content. Client quality gate and 150 browser tests pass. A fresh backend served the exact dist bytes for the shell, entry script, node library, and node search popup. No backend execution changes or live model-generation claim.
Publish the validated MoDiff-client 19eeebf build and document that the saved resourceMode is execution policy, independent of authoring preference. Preserve backend-owned user fields and installed local Gallery content. Validation: paired client quality gate (1,112 Node tests) and browser gate (2 shared-control and 150 Studio tests). A fresh backend served exact build bytes on the shell and four JavaScript routes, including the new controls. Backend Python execution code is unchanged; no live model reuse claim.
Expose ordinary graph drafts, required component wiring and shared generator inputs. Bundle the matching stage authoring client and refresh bundle evidence without promoting model qualification. Validated with the full backend gate, optional-runtime contracts, native UI journeys and cached SDXL execution.
Use resident float32 defaults for unconditional image starters, clear unused conditioning artifacts, and mark loader overrides optional. Refresh dependent catalog metadata and the paired frontend bundle; preserve compiled graph semantics and model pins. Correct Windows fixture encoding/Git blob checks and forked-state decode assertions. Base pytest: 3163 passed, 500 skipped; verified optional runtime: 3703 passed, 12 skipped. Client quality and browser gates passed; scoped native image/audio/video runs are recorded in the mirrored plan. Gallery/release qualification gaps remain open.
Review paired image contracts, native variants, custom nodes, and bounded resource recovery. Include the matching client bundle and Windows demo guide. Base tests: 3680 passed; reviewed optional runtime: 4264 passed. Generated media and model weights remain outside Git.
The generated readiness ledger was hidden by the data directory ignore rule. Track its deterministic source inventory explicitly. Read the UTF-8 history fixture with the same encoding as the application, independent of the Windows system locale.
Ship image/audio authoring contracts, structured Python imports, examples and matching client assets. Preserve explicitly removed dynamic Guidance controls and their effective values across persistence.
Align exact snapshot loading with managed cache discovery, preserve native ACE embedding and VAE entry points, and correct isolated Windows backend startup. Include focused regressions, the verified client bundle and hash-only inventory refresh. Consolidate engineering lessons and acceptance requirements; preserve the composite contract while retiring historical plans. Matching MoDiff-client commit: 4a100d2fcd33e4fbba6160d3e1080c0158d397c8. Validation retained: full backend 3727 passed, 536 skipped; optional runtime 138 passed; production SSD and sequential audio generation/playback passed. Latest focused review: 73 passed, 1 skipped, 264 subtests. No additional inference for documentation cleanup.
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
1 participant
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.
Summary
Scope And Human Review
none:Checks
uvx --from ruff==0.12.7 ruff check . --select E9,Fuv pip check --python .venv/bin/python(use.venv/Scripts/python.exeon Windows)./.venv/bin/python -m modiff.preflight --json --check-port 8088 --fail-on-error./.venv/bin/python -m pytest -qbash -n run.shwhen a POSIX shell is availableValidation Results
not claimed:Compatibility, Security, And Proof