perf: use cached file for wheel inspection #7916
Merged
Conversation
This file contains 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
ralbertazzi
force-pushed
the
perf/use-cached-weel-for-inspection
branch
from
14329f9
to
176b109
Compare
dimbleby
reviewed
May 13, 2023
ralbertazzi
force-pushed
the
perf/use-cached-weel-for-inspection
branch
2 times, most recently
from
1dda952
to
45617e1
Compare
radoering
reviewed
May 13, 2023
ralbertazzi
force-pushed
the
perf/use-cached-weel-for-inspection
branch
2 times, most recently
from
7aeb909
to
48df4ab
Compare
Closed
ralbertazzi
force-pushed
the
perf/use-cached-weel-for-inspection
branch
3 times, most recently
from
8570d06
to
06f8dd6
Compare
radoering
reviewed
May 15, 2023
ralbertazzi
force-pushed
the
perf/use-cached-weel-for-inspection
branch
from
06f8dd6
to
9d68507
Compare
radoering
approved these changes
May 15, 2023
mwalbeck
pushed a commit
to mwalbeck/docker-python-poetry
that referenced
this pull request
May 23, 2023
This PR contains the following updates: | Package | Update | Change | |---|---|---| | [poetry](https://python-poetry.org/) ([source](https://github.com/python-poetry/poetry), [changelog](https://python-poetry.org/history/)) | minor | `1.4.2` -> `1.5.0` | --- ### Release Notes <details> <summary>python-poetry/poetry</summary> ### [`v1.5.0`](https://github.com/python-poetry/poetry/blob/HEAD/CHANGELOG.md#​150---2023-05-19) [Compare Source](python-poetry/poetry@1.4.2...1.5.0) ##### Added - **Introduce the new source priorities `explicit` and `supplemental`** ([#​7658](python-poetry/poetry#7658), [#​6879](python-poetry/poetry#6879)). - **Introduce the option to configure the priority of the implicit PyPI source** ([#​7801](python-poetry/poetry#7801)). - Add handling for corrupt cache files ([#​7453](python-poetry/poetry#7453)). - Improve caching of URL and git dependencies ([#​7693](python-poetry/poetry#7693), [#​7473](python-poetry/poetry#7473)). - Add option to skip installing directory dependencies ([#​6845](python-poetry/poetry#6845), [#​7923](python-poetry/poetry#7923)). - Add `--executable` option to `poetry env info` ([#​7547](python-poetry/poetry#7547)). - Add `--top-level` option to `poetry show` ([#​7415](python-poetry/poetry#7415)). - Add `--lock` option to `poetry remove` ([#​7917](python-poetry/poetry#7917)). - Add experimental `POETRY_REQUESTS_TIMEOUT` option ([#​7081](python-poetry/poetry#7081)). - Improve performance of wheel inspection by avoiding unnecessary file copy operations ([#​7916](python-poetry/poetry#7916)). ##### Changed - **Remove the old deprecated installer and the corresponding setting `experimental.new-installer`** ([#​7356](python-poetry/poetry#7356)). - **Introduce `priority` key for sources and deprecate flags `default` and `secondary`** ([#​7658](python-poetry/poetry#7658)). - Deprecate `poetry run <entry point>` if the entry point was not previously installed via `poetry install` ([#​7606](python-poetry/poetry#7606)). - Only write the lock file if the installation succeeds ([#​7498](python-poetry/poetry#7498)). - Do not write the unused package category into the lock file ([#​7637](python-poetry/poetry#7637)). ##### Fixed - Fix an issue where Poetry's internal pyproject.toml continually grows larger with empty lines ([#​7705](python-poetry/poetry#7705)). - Fix an issue where Poetry crashes due to corrupt cache files ([#​7453](python-poetry/poetry#7453)). - Fix an issue where the `Retry-After` in HTTP responses was not respected and retries were handled inconsistently ([#​7072](python-poetry/poetry#7072)). - Fix an issue where Poetry silently ignored invalid groups ([#​7529](python-poetry/poetry#7529)). - Fix an issue where Poetry does not find a compatible Python version if not given explicitly ([#​7771](python-poetry/poetry#7771)). - Fix an issue where the `direct_url.json` of an editable install from a git dependency was invalid ([#​7473](python-poetry/poetry#7473)). - Fix an issue where error messages from build backends were not decoded correctly ([#​7781](python-poetry/poetry#7781)). - Fix an infinite loop when adding certain dependencies ([#​7405](python-poetry/poetry#7405)). - Fix an issue where pre-commit hooks skip pyproject.toml files in subdirectories ([#​7239](python-poetry/poetry#7239)). - Fix an issue where pre-commit hooks do not use the expected Python version ([#​6989](python-poetry/poetry#6989)). - Fix an issue where an unclear error message is printed if the project name is the same as one of its dependencies ([#​7757](python-poetry/poetry#7757)). - Fix an issue where `poetry install` returns a zero exit status even though the build script failed ([#​7812](python-poetry/poetry#7812)). - Fix an issue where an existing `.venv` was not used if `in-project` was not set ([#​7792](python-poetry/poetry#7792)). - Fix an issue where multiple extras passed to `poetry add` were not parsed correctly ([#​7836](python-poetry/poetry#7836)). - Fix an issue where `poetry shell` did not send a newline to `fish` ([#​7884](python-poetry/poetry#7884)). - Fix an issue where `poetry update --lock` printed operations that were not executed ([#​7915](python-poetry/poetry#7915)). - Fix an issue where `poetry add --lock` did perform a full update of all dependencies ([#​7920](python-poetry/poetry#7920)). - Fix an issue where `poetry shell` did not work with `nushell` ([#​7919](python-poetry/poetry#7919)). - Fix an issue where subprocess calls failed on Python 3.7 ([#​7932](python-poetry/poetry#7932)). - Fix an issue where keyring was called even though the password was stored in an environment variable ([#​7928](python-poetry/poetry#7928)). ##### Docs - Add information about what to use instead of `--dev` ([#​7647](python-poetry/poetry#7647)). - Promote semantic versioning less aggressively ([#​7517](python-poetry/poetry#7517)). - Explain Poetry's own versioning scheme in the FAQ ([#​7517](python-poetry/poetry#7517)). - Update documentation for configuration with environment variables ([#​6711](python-poetry/poetry#6711)). - Add details how to disable the virtualenv prompt ([#​7874](python-poetry/poetry#7874)). - Improve documentation on whether to commit `poetry.lock` ([#​7506](python-poetry/poetry#7506)). - Improve documentation of `virtualenv.create` ([#​7608](python-poetry/poetry#7608)). ##### poetry-core ([`1.6.0`](https://github.com/python-poetry/poetry-core/releases/tag/1.6.0)) - Improve error message for invalid markers ([#​569](python-poetry/poetry-core#569)). - Increase robustness when deleting temporary directories on Windows ([#​460](python-poetry/poetry-core#460)). - Replace `tomlkit` with `tomli`, which changes the interface of some *internal* classes ([#​483](python-poetry/poetry-core#483)). - Deprecate `Package.category` ([#​561](python-poetry/poetry-core#561)). - Fix a performance regression in marker handling ([#​568](python-poetry/poetry-core#568)). - Fix an issue where wildcard version constraints were not handled correctly ([#​402](python-poetry/poetry-core#402)). - Fix an issue where `poetry build` created duplicate Python classifiers if they were specified manually ([#​578](python-poetry/poetry-core#578)). - Fix an issue where local versions where not handled correctly ([#​579](python-poetry/poetry-core#579)). </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 PR becomes conflicted, or you tick the rebase/retry checkbox. 🔕 **Ignore**: Close this PR and you won't be reminded about this update again. --- - [ ] <!-- rebase-check -->If you want to rebase/retry this PR, check this box --- This PR has been generated by [Renovate Bot](https://github.com/renovatebot/renovate). <!--renovate-debug:eyJjcmVhdGVkSW5WZXIiOiIzNS44Mi4wIiwidXBkYXRlZEluVmVyIjoiMzUuODIuMCIsInRhcmdldEJyYW5jaCI6Im1hc3RlciJ9--> Reviewed-on: https://git.walbeck.it/walbeck-it/docker-python-poetry/pulls/717 Co-authored-by: renovate-bot <[email protected]> Co-committed-by: renovate-bot <[email protected]>
ralbertazzi
added a commit
to ralbertazzi/poetry
that referenced
this pull request
May 24, 2023
radoering
pushed a commit
that referenced
this pull request
May 24, 2023
|
This pull request has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs. |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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.
Another PR that goes in the way of improving the Poetry experience when dealing with huge wheels (did anyone say PyTorch? 👀 #6409 )
I'm hosting huge wheels on a repository that is not PEP 658 compliant. What happens is that Poetry will rightfully want to get package metadata from the first wheel. The package is correctly cached on filesystem in the repository cache (the whole
Authenticator+FileCacheflow) so that future downloads are not needed again.However, when a wheel is already cached on filesystem the inspection code performs an unnecessary data transfer operation from the cache location to a temporary directory. As the speed of this operation is controlled by the low
chunk_size: int = 1024value, we spend a gigantic amount of time waiting for something that we already have. Console logs also incorrectly show that a package is being downloaded, while in reality no network operation is happening.This PR fixes this behaviour by accessing the cached file directly, and relying on the download to temporary file only if the file is not cached (or the cache is disabled). In order to do so it exploits the public
url_to_file_pathutility fromcachecontrol. I thought it was best leaving any directcachecontroldependency inside theAuthenticator, which is the reason why I'm touching two classes.[ ] Updated documentation for changed code.