uv-resolver: fix conflicting extra bug during uv sync#11075
Merged
BurntSushi merged 1 commit intomainfrom Jan 29, 2025
Merged
Conversation
In #10875, I relaxed the error checking during resolution to permit dependencies like `foo[x1]`, where `x1` was defined to be conflicting. In exchange, the error was, roughly speaking, moved to installation time. This was achieved by looking at the full set of enabled extras and checking whether any conflicts occurred. If so, an error was reported. This ends up being more expressive and permits more valid configurations. However, in so doing, there was a bug in how the accumulated extras were being passed to conflict marker evaluation. Namely, we weren't accounting for the fact that if `foo[x1]` was enabled, then that fact should be carried through to all conflict marker evaluations. This is because some of those will use things like `extra != 'x1'` to indicate that it should only be included if an extra *isn't* enabled. In #10985, this manifested with PyTorch where `torch==2.4.1` and `torch==2.4.1+cpu` were being installed simultaneously. Namely, the choice to install `torch==2.4.1` was not taking into account that the `cpu` extra has been enabled. If it did, then it's conflict marker would evaluate to `false`. Since it didn't, and since `torch==2.4.1+cpu` was also being included, we ended up installing both versions. The approach I took in this PR was to add a second breadth first traversal (which comes first) over the dependency tree to accumulate all of the activated extras. Then, only in the second traversal do we actually build up the resolution graph. Unfortunately, I have no automatic regression test to include here. The regression test we _ought_ to include involves `torch`. And while we are generally find to use those in tests that only generate a lock file, the regression test here actually requires running installation. And downloading and installing `torch` in tests is bad juju. So adding a regression test for this is blocked on better infrastructure for PyTorch tests. With that said, I did manually verify that the test case in #10985 no longer installs multiple versions of `torch`. Fixes #10985
BurntSushi
force-pushed
the
ag/fix-double-torch
branch
from
9a06574 to
3c35067
Compare
charliermarsh
approved these changes
Jan 29, 2025
tmeijn
pushed a commit
to tmeijn/dotfiles
that referenced
this pull request
Feb 4, 2025
This MR contains the following updates: | Package | Update | Change | |---|---|---| | [astral-sh/uv](https://github.com/astral-sh/uv) | patch | `0.5.25` -> `0.5.27` | MR created with the help of [el-capitano/tools/renovate-bot](https://gitlab.com/el-capitano/tools/renovate-bot). **Proposed changes to behavior should be submitted there as MRs.** --- ### Release Notes <details> <summary>astral-sh/uv (astral-sh/uv)</summary> ### [`v0.5.27`](https://github.com/astral-sh/uv/blob/HEAD/CHANGELOG.md#0527) [Compare Source](astral-sh/uv@0.5.26...0.5.27) ##### Enhancements - Avoid setting permissions during tar extraction ([#​11191](astral-sh/uv#11191)) - Remove warnings for missing lower bounds ([#​11195](astral-sh/uv#11195)) - Update PubGrub to set-based outdated priority tracking ([#​11169](astral-sh/uv#11169)) - Improve error messages for `uv pip install` with `--extra` or `--all-extras` and invalid sources ([#​11193](astral-sh/uv#11193)) - Sign Docker images using GitHub attestations ([#​8685](astral-sh/uv#8685)) ##### Preview features - Don't expand self-referential extras in the build backend ([#​11142](astral-sh/uv#11142)) ##### Performance - Filter discovered Python executables by source before querying ([#​11143](astral-sh/uv#11143)) - Optimize exclusion computation for markers ([#​11158](astral-sh/uv#11158)) - Use Astral-maintained `tokio-tar` fork ([#​11174](astral-sh/uv#11174)) - Remove unneeded `.clone()` ([#​11127](astral-sh/uv#11127)) ##### Bug fixes - Fix relative paths in bytecode compilation ([#​11177](astral-sh/uv#11177)) - Percent-decode URLs in canonical comparisons ([#​11088](astral-sh/uv#11088)) - Respect concurrency limits in parallel index fetch ([#​11182](astral-sh/uv#11182)) - Use wire JSON schema for conflict items ([#​11196](astral-sh/uv#11196)) - Use explicit `_GLibCVersion` tuple in uv-python crate ([#​11122](astral-sh/uv#11122)) ##### Documentation - Add Git SHA locking behavior to docs ([#​11125](astral-sh/uv#11125)) - Add best-practice flags to `pip install` example in troubleshooting guide ([#​11194](astral-sh/uv#11194)) - Set `VIRTUAL_ENV` in Jupyter kernels ([#​11155](astral-sh/uv#11155)) - Add instructions for deactivating an environment ([#​11200](astral-sh/uv#11200)) ### [`v0.5.26`](https://github.com/astral-sh/uv/blob/HEAD/CHANGELOG.md#0526) [Compare Source](astral-sh/uv@0.5.25...0.5.26) ##### Enhancements - Add support for `uvx python` ([#​11076](astral-sh/uv#11076)) - Allow `--no-dev --invert` in `uv tree` ([#​11068](astral-sh/uv#11068)) - Update `uv python install --reinstall` to reinstall all previous versions ([#​11072](astral-sh/uv#11072)) - Consistently write log messages with capitalized first word ([#​11111](astral-sh/uv#11111)) - Suggest `--build-backend` when `--backend` is passed to `uv init` ([#​10958](astral-sh/uv#10958)) - Improve retry trace message ([#​11108](astral-sh/uv#11108)) ##### Performance - Remove unnecessary UTF-8 conversion in hash parsing ([#​11110](astral-sh/uv#11110)) ##### Bug fixes - Ignore non-hash fragments in HTML API responses ([#​11107](astral-sh/uv#11107)) - Avoid resolving symbolic links when querying Python interpreters ([#​11083](astral-sh/uv#11083)) - Avoid sharing state between universal and non-universal resolves ([#​11051](astral-sh/uv#11051)) - Error when `--script` is passing a non-PEP 723 script ([#​11118](astral-sh/uv#11118)) - Make metadata deserialization failures non-fatal in the cache ([#​11105](astral-sh/uv#11105)) - Mark metadata as dynamic when reading from built wheel cache ([#​11046](astral-sh/uv#11046)) - Propagate credentials for `<index>/simple` to `<index>/...` endpoints ([#​11074](astral-sh/uv#11074)) - Fix conflicting extra bug during `uv sync` ([#​11075](astral-sh/uv#11075)) ##### Documentation - Add PyTorch XPU instructions to the PyTorch guide ([#​11109](astral-sh/uv#11109)) - Add docs for signal handling ([#​11041](astral-sh/uv#11041)) - Explain build frontend vs. build backend ([#​11094](astral-sh/uv#11094)) - Fix formatting of `RUST_LOG` documentation ([#​10053](astral-sh/uv#10053)) - Fix typo in `--no-deps` description ([#​11073](astral-sh/uv#11073)) - Reflow CLI documentation comments ([#​11040](astral-sh/uv#11040)) - Shorten "Using existing Python versions" nav item so it fits on one line ([#​11077](astral-sh/uv#11077)) - Some minor touch-ups to the Python install guide ([#​11116](astral-sh/uv#11116)) - Update Dependabot tracking issue link ([#​11054](astral-sh/uv#11054)) - Update documentation for running in a container ([#​11052](astral-sh/uv#11052)) - Upgrade PyTorch version in documentation ([#​11114](astral-sh/uv#11114)) - Use `sys_platform` in lieu of `platform_system` in PyTorch docs ([#​11113](astral-sh/uv#11113)) - Use positive (rather than negative) markers in PyTorch examples ([#​11112](astral-sh/uv#11112)) - Fix unnecessary backslashes in brackets ([#​11059](astral-sh/uv#11059)) - Suggest setting copy link mode in GitLab integration guide ([#​11067](astral-sh/uv#11067)) </details> --- ### Configuration 📅 **Schedule**: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined). 🚦 **Automerge**: Disabled by config. Please merge this manually once you are satisfied. ♻ **Rebasing**: Whenever MR becomes conflicted, or you tick the rebase/retry checkbox. 🔕 **Ignore**: Close this MR and you won't be reminded about this update again. --- - [ ] <!-- rebase-check -->If you want to rebase/retry this MR, check this box --- This MR has been generated by [Renovate Bot](https://github.com/renovatebot/renovate). <!--renovate-debug:eyJjcmVhdGVkSW5WZXIiOiIzOS4xNDMuMCIsInVwZGF0ZWRJblZlciI6IjM5LjE1Ni4xIiwidGFyZ2V0QnJhbmNoIjoibWFpbiIsImxhYmVscyI6WyJSZW5vdmF0ZSBCb3QiXX0=-->
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
2 participants
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.
In #10875, I relaxed the error checking during resolution to permit
dependencies like
foo[x1], wherex1was defined to be conflicting.In exchange, the error was, roughly speaking, moved to installation
time. This was achieved by looking at the full set of enabled extras
and checking whether any conflicts occurred. If so, an error was
reported. This ends up being more expressive and permits more valid
configurations.
However, in so doing, there was a bug in how the accumulated extras
were being passed to conflict marker evaluation. Namely, we weren't
accounting for the fact that if
foo[x1]was enabled, then that factshould be carried through to all conflict marker evaluations. This is
because some of those will use things like
extra != 'x1'to indicatethat it should only be included if an extra isn't enabled.
In #10985, this manifested with PyTorch where
torch==2.4.1andtorch==2.4.1+cpuwere being installed simultaneously. Namely, thechoice to install
torch==2.4.1was not taking into account thatthe
cpuextra has been enabled. If it did, then it's conflictmarker would evaluate to
false. Since it didn't, and sincetorch==2.4.1+cpuwas also being included, we ended up installing bothversions.
The approach I took in this PR was to add a second breadth first
traversal (which comes first) over the dependency tree to accumulate all
of the activated extras. Then, only in the second traversal do we
actually build up the resolution graph.
Unfortunately, I have no automatic regression test to include here. The
regression test we ought to include involves
torch. And while we aregenerally find to use those in tests that only generate a lock file, the
regression test here actually requires running installation. And
downloading and installing
torchin tests is bad juju. So adding aregression test for this is blocked on better infrastructure for PyTorch
tests. With that said, I did manually verify that the test case in #10985
no longer installs multiple versions of
torch.Fixes #10985