Backport Redis 7.4 features to the 5.0 branch #3312
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
…se of empty command stack (redis#3068)
redis#3001) * Update repr of important classes with module name and recommended "< ... >" syntax. * update tests which examine repr * formatting
…bled flag (redis#3158) * change if the cluster-mode is enabled by trying run CLUSTER SLOT insted of INFO * fix typo * fixing cluster mode is not enabled on this node tests * remove changes on asyncio * rename mock flag to be more consistent * optimizing async cluster creation using CLUSTER SLOT command instead of INFO command * fixing test. Before INFO and CLUSTER_SLOT was used for performing the connection, now only the CLUSTER_SLOT, so the total commands is minus 1 * remove dot at the end of string * remove unecessary print from test * fix lint problems --------- Co-authored-by: Willian Moreira <[email protected]>
clarified wording + one case of deprecated .close() changed to .aclose()
Issue redis#3154 The CLIENT KILL command has a new option, called MAXAGE, to kill clients older than a given age. Add support for this new option. Co-authored-by: Gabriel Erzse <[email protected]>
* Support NOVALUES parameter for HSCAN Issue redis#3153 The NOVALUES parameter instructs HSCAN to only return the hash keys, without values. Co-authored-by: Gabriel Erzse <[email protected]>
The CNs in the test certificates (CA, server and client) were all the same, which is not right. Make them different. Cleanup the Jupyter notebook that documents how SSL connections work. Make the code blocks actually run, by using the test certificates from devenv. Simplify a bit the OCSP examples, since they can't run anyway, due to lack of OCSP in our test infrastructure, so they are rather informative. Co-authored-by: Gabriel Erzse <[email protected]>
XREAD now supports reading the last message in the stream, with the + special Id. The code already supports this, so just add some documentation to an existing Jupyter notebook. Fixes redis#3155 Co-authored-by: Gabriel Erzse <[email protected]>
Adopt the Google docstrings style as the preferred way to document code. Update one of the existing docstrings to the new format, to showcase how it would look like. Enable the `napoleon` preprocessor in Sphinx, to be able to process Google style docstrings. Take the opportunity to make the copyright year dynamic, so we don't have to manually keep it up to date. --------- Co-authored-by: Gabriel Erzse <[email protected]>
Pyright treats explicit and implicit `Any` differently, with implicit `Any` being treated as `Unknown`. Pyright raised a warning indicating an unknown member type for 'Awaitable' when used in 'ResponseT'. Using the `incr` method as an example, the warning is: ``` "warning: Type of 'incr' is partially unknown Type of 'incr' is '(name: bytes | str | memoryview, amount: int = 1) -> (Awaitable[Unknown] | Any)' (reportUnknownMemberType)" ``` By explicitly specifying 'Awaitable[Any]' in the union for 'ResponseT', this resolves the ambiguity about the member type.
Use the new logo, and use the SVG version of the logo in all places, including the README. Replace most of the redis.com URLs with their new redis.io counterparts. Co-authored-by: Gabriel Erzse <[email protected]>
The address_remap callback received a tuple with the host and the port. Make this clear in the typing hints.
Make the parsing of memory stats response more robust, to not break on changes that will be added to the Redis server. Also make a test related to client kill by maxage more resilient. Co-authored-by: Gabriel Erzse <[email protected]>
Add type hints to backoff.py
Adapt the tests to use the standard Redis docker image where possible, instead of using the Redis Stack image in all places. This way we can run the CI against different versions of Redis and Redis Stack. Run the test cluster based on the default Redis docker image, i.e. without Stack support. This is needed in order to make the CI pass until Stack gets a new release. What is needed is to mark all Stack related tests accordingly, and then they don't get executed in the cluster tests (they are already filtered out in `invoke cluster-tests`). Take the opportunity to remove some empty tests and to fix the naming in two tests. --------- Co-authored-by: Gabriel Erzse <[email protected]>
Support hash field expiration commands that become available with Redis 7.4. Adapt some tests to match recent server-side changes. Update tests related to memory stats. Make CLIENT KILL test not run with cluster. Disable tests related to Graph module. The Graph module is no longer part of Redis Stack, so for the moment disable all related tests. --------- Co-authored-by: Gabriel Erzse <[email protected]>
Support timeseries insertion filters for samples that are close to each other in time and value. Use the documented way to disable compression, i.e. `ENCODING UNCOMPRESSED` instead of `UNCOMPRESSED`. Polish the documentation related to timeseries. Align things needed around CI, to make sure all tests are actually executed. BREAKING CHANGES: 1. Remove the `uncompressed` flag from TS.ALTER, since compression of existing timeseries cannot be changed. This should not have been used, so there should be no real impact. 2. For the TS.ADD command (TimeSeriesCommands.add method): the `duplicate_policy` Python parameter that was mapping to `ON DUPLICATE` was now rewired to map to `DUPLICATE POLICY`. A new Python parameter called `on_duplicate` was added, that maps to `ON DUPLICATE`. The expected impact of this change is low.
Add more tests for GEO search, to cover the query operators `within`, `contains`, `intersects` and `disjoint`, for POINT and POLYGON, i.e. the currently supported shapes and operators.
Add support for indexing and searching missing and empty values. Currently there are some limitation from the server side, for example empty values are supported only for TEXT and TAG fields.
With RediSearch dialect 5 less special characters need escaping when searching, given that TAG fields are used. Add a test to show some examples.
…#3260) Bumps [rojopolis/spellcheck-github-actions](https://github.com/rojopolis/spellcheck-github-actions) from 0.36.0 to 0.37.0. - [Release notes](https://github.com/rojopolis/spellcheck-github-actions/releases) - [Changelog](https://github.com/rojopolis/spellcheck-github-actions/blob/master/CHANGELOG.md) - [Commits](rojopolis/spellcheck-github-actions@0.36.0...0.37.0) --- updated-dependencies: - dependency-name: rojopolis/spellcheck-github-actions dependency-type: direct:production update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Fix some some statements with missing asserts.
Bumps [actions/stale](https://github.com/actions/stale) from 3 to 9. - [Release notes](https://github.com/actions/stale/releases) - [Changelog](https://github.com/actions/stale/blob/main/CHANGELOG.md) - [Commits](actions/stale@v3...v9) --- updated-dependencies: - dependency-name: actions/stale dependency-type: direct:production update-type: version-update:semver-major ... Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Make the parsing of XINFO STREAM response more resilient, by handling the case when first and last entries are None after XTRIM. Improve the parsing of consumers related info.
The CI tests were not running with RESP3 protocol, it was just an illusion that they do. Fix this, and also preserve coverage and test artifacts from those runs too. Some issues have surfaced after the change. The most notable issue is a bug in hiredis-py, which prevents it from being used in cluster mode at least. Make sure cluster tests do not run with hiredis-py. Also make sure some specific unit tests do not run with hiredis-py. One other issue with hiredis-py is fixed in this commit. Use a sentinel object instance to signal lack of data in hiredis-py, instead of piggybacking of `False`, which can also be returned by parsing valid RESP payloads. Some of the unit tests, mostly for modules, were failing, they are now updated so that they pass. Remove async parser from test fixture params. Leave the decision for the async parser to be used in tests to be taken based on the availability of hiredis-py, and on the protocol that is set for the tests. Otherwise when hiredis-py is available we would also run the non-hiredis tests.
We did not push code coverage from the master branch since a while. Try to add a Codecov token, to get back on track.
Try to not restrict the versions of pytest dependencies, see if we can get rid of the segmentation faults that we see pretty often in the CI. Take the opportunity to sort the dev_requirements.txt file.
Although the Graph module was deprecated, we want to have tests running as long as we keep the client code. Therefore re-enable the Graph tests, by adding to the docker-compose stack an older redis-stack-server image. The Graph tests run only with RESP2. Tweak the CI so that it runs against the latest RC release for now. This way we get all the features in RC release of server, so we can execute all tests, including the ones for hash field expiration. Some test failures brought to light issues with the RESP3 push message invalidation handler, align a bit the code in that area. Some more housekeeping around tests, here and there.
…#3301) Bumps [rojopolis/spellcheck-github-actions](https://github.com/rojopolis/spellcheck-github-actions) from 0.37.0 to 0.38.0. - [Release notes](https://github.com/rojopolis/spellcheck-github-actions/releases) - [Changelog](https://github.com/rojopolis/spellcheck-github-actions/blob/master/CHANGELOG.md) - [Commits](rojopolis/spellcheck-github-actions@0.37.0...0.38.0) --- updated-dependencies: - dependency-name: rojopolis/spellcheck-github-actions dependency-type: direct:production update-type: version-update:semver-minor ... Signed-off-by: dependabot[bot] <[email protected]> Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Add missing type hints in the retry.py file and related tests.
For asyncio connection errors, include the details in the error message, instead of only including the error code. Co-authored-by: dmitry.kanev <[email protected]> Co-authored-by: Gabriel Erzse <[email protected]>
Adapt the code and some of the tests to match the changes done in the Redis 7.4 RC2 release.
Connection errors are formatted in four places, sync and async, network socket and unix socket. Each place has some small differences compared to the others, while they could be, and should be, formatted in an uniform way. Factor out the logic in a helper method and call that method in all four places. Arguably we lose some specificity, e.g. the words "unix socket" won't be there anymore, but it is more valuable to not have code duplication.
…version) (redis#3165) * fix disconnects * skip test in cluster * add test * save return value from handle_push_response (without it 'read_response' return the push message) * insert return response from cache to the try block to prevent connection leak * enable to get connection with data avaliable to read in csc mode and change can_read_destructive to not read data * fix check if socket is empty (at_eof() can return False but this doesn't mean there's definitely more data to read) --------- Co-authored-by: Chayim <[email protected]>
Make it possible to configure at field level how search results are decoded. Fixes: redis#2772, redis#2275
gerzse
force-pushed
the
backport-redis74-features
branch
from
8c79060 to
62fc873
Compare
vladvildanov
approved these changes
Jul 11, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Pull Request check-list
Please make sure to review and check all of these items:
NOTE: these things are not required to open a PR and can be done
afterwards / while the PR is open.
Description of change
The goal is to bring everything except client side caching from master to the 5.0 branch, so we can release a non-beta version from there.