Bump ComfyUI to v0.15.0 and frontend to v1.39.16#1614
Conversation
dosubot
bot
added
size:L
📝 WalkthroughWalkthroughCompiled requirements for macOS and Windows were updated to bump multiple comfyui-related packages and add OpenGL-related packages; frontend and ComfyUI version fields were incremented and core requirements patch adjusted accordingly. Changes
Possibly related PRs
Suggested reviewers
Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
coderabbitai
bot
requested review from
christian-byrne and
luke-mino-altherr
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
🧹 Nitpick comments (1)
assets/requirements/macos.compiled (1)
2-4: macOS compile command includes--extra-index-url https://download.pytorch.org/whl/cu130.The CUDA 13.0 PyTorch index is unnecessary for
aarch64-apple-darwinand is the reasonmarkupsafegets pulled as 3.0.2 from the PyTorch index instead of 3.0.3 from PyPI (Lines 155-159). Since this is autogenerated, the compile script likely uses a shared template. Not a functional issue — uv won't resolve macOS ARM wheels from the cu130 index for CUDA packages — but removing it for the macOS compile would produce a cleaner result.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@assets/requirements/macos.compiled` around lines 2 - 4, Remove the unnecessary CUDA PyTorch index from the macOS compile invocation: in the autogenerated macOS compile command that targets "--python-platform aarch64-apple-darwin" (the commented "uv pip compile ..." invocation) drop the flag "--extra-index-url https://download.pytorch.org/whl/cu130" so the macOS artifact doesn't pick up CUDA-specific indices (which cause markupsafe to be pulled as 3.0.2); update the template or generation logic that emits that compile line so it only adds the cu130 extra-index-url for CUDA-capable platforms, not for aarch64-apple-darwin.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Nitpick comments:
In `@assets/requirements/macos.compiled`:
- Around line 2-4: Remove the unnecessary CUDA PyTorch index from the macOS
compile invocation: in the autogenerated macOS compile command that targets
"--python-platform aarch64-apple-darwin" (the commented "uv pip compile ..."
invocation) drop the flag "--extra-index-url
https://download.pytorch.org/whl/cu130" so the macOS artifact doesn't pick up
CUDA-specific indices (which cause markupsafe to be pulled as 3.0.2); update the
template or generation logic that emits that compile line so it only adds the
cu130 extra-index-url for CUDA-capable platforms, not for aarch64-apple-darwin.
ℹ️ Review info
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (6)
assets/requirements/macos.compiledassets/requirements/windows_amd.compiledassets/requirements/windows_cpu.compiledassets/requirements/windows_nvidia.compiledpackage.jsonscripts/core-requirements.patch
There was a problem hiding this comment.
Pull request overview
Updates the embedded ComfyUI backend and pinned frontend package versions, and refreshes the platform-specific compiled Python requirement lockfiles to match the new upstream ComfyUI release.
Changes:
- Bump
package.jsonconfig versions: ComfyUI0.14.2 → 0.15.0, frontend1.38.14 → 1.39.16. - Update
scripts/core-requirements.patchto remove the new upstreamcomfyui-frontend-package==1.39.16pin and match updated template/docs pins. - Regenerate compiled requirements for macOS and Windows variants, including new deps (e.g.,
glfw,pyopengl,pyopengl-accelerate) and updated comfy-related packages.
Reviewed changes
Copilot reviewed 6 out of 6 changed files in this pull request and generated 4 comments.
Show a summary per file
| File | Description |
|---|---|
scripts/core-requirements.patch |
Updates the patch applied to ComfyUI’s requirements.txt for v0.15.0 (removing the new frontend pin and updating context versions). |
package.json |
Bumps configured ComfyUI and frontend versions used by repo tooling/automation. |
assets/requirements/windows_nvidia.compiled |
Updated Windows NVIDIA compiled requirements for the new ComfyUI release (new/updated pins and deps). |
assets/requirements/windows_cpu.compiled |
Updated Windows CPU compiled requirements for the new ComfyUI release (new/updated pins and deps). |
assets/requirements/windows_amd.compiled |
Updated Windows AMD compiled requirements for the new ComfyUI release (new/updated pins and deps). |
assets/requirements/macos.compiled |
Updated macOS compiled requirements for the new ComfyUI release (new/updated pins and deps). |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| glfw==2.10.0 | ||
| # via -r assets/ComfyUI/requirements.txt | ||
| # from https://pypi.org/simple |
There was a problem hiding this comment.
The header of this regenerated macOS .compiled file appears inconsistent with the repo’s automation: it lists an --extra-index-url (cu130) in the file header even though the embedded generation command doesn’t include it, and the update_compiled_requirements.yml workflow expects a different uv pip compile command entirely. This will lead to workflow validation warnings and makes the file hard to reproduce reliably. Please regenerate with the exact workflow command, or update the workflow to match the header so they stay in sync.
| pyopengl==3.1.10 | ||
| # via -r assets/ComfyUI/requirements.txt | ||
| # from https://pypi.org/simple | ||
| pyopengl-accelerate==3.1.10 | ||
| # via -r assets/ComfyUI/requirements.txt |
There was a problem hiding this comment.
PR description says macOS torchaudio was confirmed as 2.9.1, but assets/requirements/macos.compiled currently pins torchaudio==2.10.0 (see line ~310). Please either update the compiled macOS requirements to the intended torchaudio version or adjust the PR description so it matches what will be shipped.
| # from https://download.pytorch.org/whl/cu130 | ||
| comfy-aimdo==0.1.8 | ||
| comfy-aimdo==0.2.1 | ||
| # via -r assets/ComfyUI/requirements.txt | ||
| # from https://pypi.org/simple |
There was a problem hiding this comment.
The header in this regenerated .compiled file indicates it was produced with a different uv pip compile command (inputs/flags and cu130 extra-index) than the one configured in .github/workflows/update_compiled_requirements.yml (which is also used to validate that line 2 matches). This will cause the workflow’s validation step to warn and may lead to future regenerations producing different lockfiles. Please align the workflow env commands with the actual generation command (or regenerate this file using the workflow’s command set) so they match exactly.
| # click | ||
| # tqdm | ||
| # from https://pypi.org/simple | ||
| comfy-aimdo==0.1.8 | ||
| comfy-aimdo==0.2.1 | ||
| # via -r assets/ComfyUI/requirements.txt |
There was a problem hiding this comment.
This regenerated .compiled file’s header embeds a specific uv pip compile command (including the manager_requirements input and explicit python platform/version flags) that does not match the command configured in .github/workflows/update_compiled_requirements.yml. That mismatch triggers the workflow’s validation warning and makes it harder to reproduce/regenerate these files consistently. Please update either the workflow’s env command or the generation process so the header command and workflow command match exactly.
There was a problem hiding this comment.
🧹 Nitpick comments (1)
assets/requirements/macos.compiled (1)
112-114: OpenGL packages on macOS: wheel availability confirmed, deprecation caveat noted.PyPI lists a
pyopengl_accelerate-3.1.10-cp312-cp312-macosx_11_0_arm64.whl, so installation on the targetaarch64-apple-darwin/ Python 3.12 platform will succeed without a source build.That said, Apple deprecated OpenGL in macOS 10.14 (Mojave). At runtime, GLFW and PyOpenGL will still work via the legacy compatibility layer, but users may see OS-level deprecation warnings. If this is an optional/headless path in ComfyUI, consider verifying that the relevant code guards its OpenGL usage appropriately (e.g., feature-flag or try/except) so macOS users without a working OpenGL driver don't hit hard failures.
Also applies to: 228-233
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@assets/requirements/macos.compiled` around lines 112 - 114, The requirements list includes glfw (glfw==2.10.0) and PyOpenGL; because macOS deprecates OpenGL, update the runtime code that touches OpenGL (e.g., functions like initialize_glfw, create_gl_context, or any module that imports glfw/PyOpenGL) to guard usage with a feature flag or robust try/except: detect platform/driver availability, wrap imports and context creation in try/except, log a clear non-fatal warning on failure and fall back to the headless/optional path instead of raising, and ensure any code paths that assume a working GL context check the flag before calling GL functions.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Nitpick comments:
In `@assets/requirements/macos.compiled`:
- Around line 112-114: The requirements list includes glfw (glfw==2.10.0) and
PyOpenGL; because macOS deprecates OpenGL, update the runtime code that touches
OpenGL (e.g., functions like initialize_glfw, create_gl_context, or any module
that imports glfw/PyOpenGL) to guard usage with a feature flag or robust
try/except: detect platform/driver availability, wrap imports and context
creation in try/except, log a clear non-fatal warning on failure and fall back
to the headless/optional path instead of raising, and ensure any code paths that
assume a working GL context check the flag before calling GL functions.
luke-mino-altherr
left a comment
luke-mino-altherr
left a comment
There was a problem hiding this comment.
The dependency versions are slightly higher than what we set in cloud: https://github.com/Comfy-Org/cloud/pull/2554/changes
Not sure why, but the slight patch bumps seem fine.
Summary
package.jsonconfig.comfyUI.versionfrom0.14.2to0.15.0package.jsonconfig.frontend.versionfrom1.38.14to1.39.16(matching ComfyUIv0.15.0requirements.txtcomfyui-frontend-packagepin)scripts/core-requirements.patchto match ComfyUIv0.15.0requirements.txt┆Issue is synchronized with this Notion page by Unito
Summary by CodeRabbit