Promote dev → main: unified DFlash execution and PyPI versioning - #175
Draft
drunkcoding wants to merge 2 commits into
Draft
Promote dev → main: unified DFlash execution and PyPI versioning#175drunkcoding wants to merge 2 commits into
drunkcoding wants to merge 2 commits into
Conversation
#174) * build: derive package version from git tags via setuptools-scm Replace the env-var/hardcoded version (MOEINF_VERSION default 0.0.1) with setuptools-scm. The resolved version is written to moe_infinity/_version.py and baked into the sdist PKG-INFO, so it survives a source reinstall instead of reverting to the setup.py fallback. * build: git-ignore generated _version.py and prune sdist contents setuptools-scm regenerates moe_infinity/_version.py on every build, and its sdist file-finder includes all tracked files; prune docs/benchmarks/tests/examples/docker/scripts to keep the published sdist lean. * ci(publish): version stable and nightly PyPI builds from git tags Stable builds now version from the pushed tag instead of always publishing 0.0.1; nightly drops the setup-version job and NIGHTLY_BASE_VERSION and lets setuptools-scm emit X.Y.(Z+1).devN. Both checkouts use fetch-depth 0 and install setuptools-scm. * ci: install setuptools-scm and fetch tags in PR build jobs PR sdist and editable-install jobs need setuptools-scm present and full history (fetch-depth 0) so the version resolves and _version.py generation is validated. * docs: document setuptools-scm versioning and drop manual version steps Remove the manual __init__.py, setup.py, and NIGHTLY_BASE_VERSION bump steps, document tag-driven versioning and the initial v0.0.1 anchor tag, add setuptools-scm to source build deps, and record the fix in the changelog. * ci(publish): authenticate to PyPI via trusted publishing (OIDC) The pypa/gh-action-pypi-publish action reads credentials only from its 'user'/'password' inputs, never from TWINE_USERNAME/TWINE_PASSWORD env vars, so the existing env block was dead config and every nightly publish fell back to OIDC and failed with 'invalid-publisher'. Drop the dead env and add id-token: write (missing entirely from publish.yml) so trusted publishing works once a PyPI publisher is registered for this repo + workflow. * docs(release): document PyPI trusted-publishing setup Referenced by the publish workflows; lists the exact GitHub trusted-publisher registration values (owner/repo/workflow) required on PyPI before the first publish. --------- Co-authored-by: drunkcoding <leyang.xue@ed.ac.uk>
) * feat(spec_decode): add unified execution protocols Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * refactor(dflash): unify singleton session execution Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * feat(spec_decode): add capability-selected session driver Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * feat(dflash): add batched bare-HF execution Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * feat(dflash): add row-aware rich backend Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * test(dflash): guard legacy batch-loop retirement Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * feat(mla): add engine-owned DeepSeek paged cache Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * feat(runtime): add row-aware rich forward Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * feat(api): migrate deprecated DFlash generation facade Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * feat(serving): add speculative session cache contexts Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * fix(serving): preserve speculative scheduler groups Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * feat(serving): integrate persistent DFlash sessions Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * feat(api): expose paged MLA admission controls Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * feat(dflash): separate pairing and executor evidence Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * test(serving): isolate standalone module loading Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * bench(dflash): add unified execution validation Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * docs(dflash): document unified execution behavior Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * docs(serving): document speculative cache modes Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * docs: update DFlash architecture compatibility Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * docs: record unified DFlash changes Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * docs(dflash): add unified execution plans Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * test(dflash): enforce compatibility documentation Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * test: remove deleted Snowflake fixture path Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * fix(api): preserve partial startup namespaces Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * ci(dflash): run unified validation end to end Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> * fix(serving): preserve cleanup metadata on Python 3.10 Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent) Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai> --------- Co-authored-by: drunkcoding <leyang.xue@ed.ac.uk> Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
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
Status
Draft/WIP pending review and CI validation.