GH-40342: [C++] move LocalFileSystem to the registry #40356
Merged
Conversation
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
github-actions
bot
added
Component: R
Component: C++
Component: Python
Component: FlightRPC
Component: Documentation
awaiting committer review
bkietz
force-pushed
the
40342-localfs-module
branch
from
ca25b8e to
1f5e319
Compare
bkietz
force-pushed
the
40342-localfs-module
branch
from
1f5e319 to
7e74b80
Compare
bkietz
force-pushed
the
40342-localfs-module
branch
2 times, most recently
from
fd4de2a to
211b82e
Compare
Member
jorisvandenbossche
left a comment
jorisvandenbossche
left a comment
There was a problem hiding this comment.
Python changes look good to me
github-actions
bot
added
awaiting changes
bkietz
force-pushed
the
40342-localfs-module
branch
from
211b82e to
320e4d5
Compare
github-actions
bot
added
awaiting change review
bkietz
force-pushed
the
40342-localfs-module
branch
2 times, most recently
from
be4468a to
46758bc
Compare
felipecrv
reviewed
Mar 22, 2024
cpp/src/arrow/filesystem/localfs.cc
Outdated
Contributor
There was a problem hiding this comment.
Is this really necessary? The local scheme?
Member
Author
There was a problem hiding this comment.
It seemed more inconsistent to refuse the scheme since the type_name() of other filesystems is also a usable scheme. I can remove it
github-actions
bot
added
awaiting changes
felipecrv
approved these changes
Mar 22, 2024
github-actions
bot
added
awaiting change review
Member
Author
|
The only failures are a known flake #40675 |
felipecrv
approved these changes
Apr 23, 2024
github-actions
bot
added
awaiting merge
bkietz
changed the title
zanmato1984
pushed a commit
to zanmato1984/arrow
that referenced
this pull request
Apr 24, 2024
) ### Rationale for this change Moving LocalFileSystem into the registry is a good first usage and will help us hammer out which aspects of built in file systems should remain public. ### What changes are included in this PR? A factory for LocalFileSystem is added to the registry. `FileSystem::MakeUri` ( apache#18316 ) is added to enable roundtripping filesystems through URIs. `file://` URIs now support a use_mmap query parameter, and `local://` URIs are also supported as an alias. <h6 id="reduce">Reducing the set of bound classes</h6> Some unnecessary bindings to the LocalFileSystem concrete class are removed. This demonstrates that with a registered factory pattern, it is no longer necessary to keep a class hierarchy public for binding. Eventually (if desired), we can move concrete subclasses of FileSystem entirely out of public headers. ### Are these changes tested? Yes, all existing tests for file:// URIs continue to pass ### Are there any user-facing changes? For consistency, local:// URIs will now be considered equivalent to corresponding file:// URIs * GitHub Issue: apache#40342 Authored-by: Benjamin Kietzman <[email protected]> Signed-off-by: Benjamin Kietzman <[email protected]>
|
After merging your PR, Conbench analyzed the 7 benchmarking runs that have been run so far on merge-commit fd75cbd. There were no benchmark performance regressions. 🎉 The full Conbench report has more details. It also includes information about 45 possible false positives for unstable benchmarks that are known to sometimes produce them. |
Comment on lines
-1113
to
+1115
| @staticmethod | ||
| @binding(True) # Required for cython < 3 | ||
| def _reconstruct(kwargs): | ||
| # __reduce__ doesn't allow passing named arguments directly to the | ||
| # reconstructor, hence this wrapper. | ||
| return LocalFileSystem(**kwargs) | ||
|
|
||
| def __reduce__(self): | ||
| cdef CLocalFileSystemOptions opts = self.localfs.options() | ||
| return LocalFileSystem._reconstruct, (dict( | ||
| use_mmap=opts.use_mmap),) | ||
| uri = frombytes(GetResultValue(self.fs.MakeUri(b"/_"))) | ||
| return FileSystem._from_uri, (uri,) |
Member
There was a problem hiding this comment.
Refactoring this has broken the cython-2 build (as the removed comments indicated, that code was required for cython < 3)
(TBH, I would personally also fine with just dropping that build and requiring cython > 3 for building)
jorisvandenbossche
added a commit
that referenced
this pull request
Apr 30, 2024
Small follow-up fix for the failure introduced by #40356 * GitHub Issue: #40342 Authored-by: Joris Van den Bossche <[email protected]> Signed-off-by: Joris Van den Bossche <[email protected]>
vibhatha
pushed a commit
to vibhatha/arrow
that referenced
this pull request
May 25, 2024
) ### Rationale for this change Moving LocalFileSystem into the registry is a good first usage and will help us hammer out which aspects of built in file systems should remain public. ### What changes are included in this PR? A factory for LocalFileSystem is added to the registry. `FileSystem::MakeUri` ( apache#18316 ) is added to enable roundtripping filesystems through URIs. `file://` URIs now support a use_mmap query parameter, and `local://` URIs are also supported as an alias. <h6 id="reduce">Reducing the set of bound classes</h6> Some unnecessary bindings to the LocalFileSystem concrete class are removed. This demonstrates that with a registered factory pattern, it is no longer necessary to keep a class hierarchy public for binding. Eventually (if desired), we can move concrete subclasses of FileSystem entirely out of public headers. ### Are these changes tested? Yes, all existing tests for file:// URIs continue to pass ### Are there any user-facing changes? For consistency, local:// URIs will now be considered equivalent to corresponding file:// URIs * GitHub Issue: apache#40342 Authored-by: Benjamin Kietzman <[email protected]> Signed-off-by: Benjamin Kietzman <[email protected]>
vibhatha
pushed a commit
to vibhatha/arrow
that referenced
this pull request
May 25, 2024
…apache#41459) Small follow-up fix for the failure introduced by apache#40356 * GitHub Issue: apache#40342 Authored-by: Joris Van den Bossche <[email protected]> Signed-off-by: Joris Van den Bossche <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
Rationale for this change
Moving LocalFileSystem into the registry is a good first usage and will help us hammer out which aspects of built in file systems should remain public.
What changes are included in this PR?
A factory for LocalFileSystem is added to the registry.
FileSystem::MakeUri( #18316 ) is added to enable roundtripping filesystems through URIs.file://URIs now support a use_mmap query parameter, andlocal://URIs are also supported as an alias.Reducing the set of bound classes
Some unnecessary bindings to the LocalFileSystem concrete class are removed. This demonstrates that with a registered factory pattern, it is no longer necessary to keep a class hierarchy public for binding. Eventually (if desired), we can move concrete subclasses of FileSystem entirely out of public headers.
Are these changes tested?
Yes, all existing tests for file:// URIs continue to pass
Are there any user-facing changes?
For consistency, local:// URIs will now be considered equivalent to corresponding file:// URIs