GH-45747: [C++] Remove deprecated ObjectType and FileStatistics, refactor hdfs code by AlenkaF · Pull Request #45998 · apache/arrow

AlenkaF · 2025-04-01T10:28:22Z

Rationale for this change

ObjectType and FileStatistics in io/hdfs.h have been deprecated for a while and can be removed.

What changes are included in this PR?

ObjectType and FileStatistics structs are removed and instead FileSystem API in arrow::fs is used. Together with this change, the hdfs connected code is moved from cpp/src/arrow/io to cpp/src/arrow/filesystem merging FileSystem and HadoopFileSystem classes from arrow::io into the public HadoopFileSystem class.

Are these changes tested?

Existing tests should pass.

Are there any user-facing changes?

Deprecated structs are removed and all hdfs related code is now a part of the filesystem module.

GitHub Issue: [C++] Remove deprecated ObjectType and FileStatistics #45747

Also closes: #22457 (not sure about io/interfaces.h?)

AlenkaF · 2025-04-01T12:56:21Z

Ah, this will not work. If we want to remove ObjectType and FileStatistics from io/hdfs.h and instead use FileSystem API, then filesystem component would have to be enabled by default. Also, if I understand correctly, we want to do a refactoring of io/hdfs anyways which would include the changes in this PR? (#22457)

cc @pitrou

pitrou · 2025-04-02T10:04:57Z

Ah, this will not work. If we want to remove ObjectType and FileStatistics from io/hdfs.h and instead use FileSystem API, then filesystem component would have to be enabled by default.

Well, ARROW_HDFS=ON could imply ARROW_FILESYSTEM=ON. I don't think that's a problem.

Also, if I understand correctly, we want to do a refactoring of io/hdfs anyways which would include the changes in this PR?

Yes, indeed. io/hdfs could be moved to filesystem/hdfs_internal or something similar.

AlenkaF · 2025-04-02T11:37:42Z

OK, I will then move io/hdfs to filesystem/hdfs_internal. Thanks!

AlenkaF · 2025-04-09T11:42:20Z

@pitrou I think this is ready for review. The failing builds have an issue opened: #46077

pitrou

Thanks for doing this @AlenkaF! Here are some comments.

pitrou · 2025-04-29T09:04:48Z

Are all these declarations actually needed by PyArrow?

No, most of them aren't and are copied from libarrow.pxd. I can remove the unused ones - but am not sure if some external application can actually use them?

pitrou · 2025-04-29T09:07:10Z

Don't we need to link to arrow::hadoop as was done above? cc @kou for advice

Hm, yeah. I will add a link as above as it makes sense.

Ah, it is there already (that explains why nothing failed =) )

arrow/cpp/src/arrow/CMakeLists.txt

Lines 857 to 861 in 53ef438

if(ARROW_HDFS)

foreach(ARROW_FILESYSTEM_TARGET ${ARROW_FILESYSTEM_TARGETS})

target_link_libraries(${ARROW_FILESYSTEM_TARGET} PRIVATE arrow::hadoop)

endforeach()

endif()

Not sure if the line with CMAKE_DL_LIBS is also needed here then?

pitrou · 2025-04-29T09:21:04Z

Ok, but we don't want to keep those two unofficial FileSystem and HadoopFileSystem classes which create confusion with the other (public) filesystem classes.

Ideally, those two classes disappear and their implementation code gets folded into the public HadoopFileSystem class.

If that's too annoying, we should at least merge those two classes and give them a less ambiguous name, for example HdfsClient.

Ok, will go with the disappearing =)
IIUC hdfs_io.h will be removed altogether:

FileSystem and HadoopFileSystem will go into hdfs.cc, folded into the public HadoopFileSystem

HdfsConnectionConfig will also go into hdfs.cc

declarations that are left will go into hdfs_internal.cc

pitrou · 2025-04-29T09:22:19Z

Most of these declarations should IMHO go into the arrow::filesystem::internal namespace, except for HdfsConnectionConfig which can go into arrow::filesystem.

AlenkaF · 2025-05-19T09:57:13Z

Hi @pitrou, could you please take a quick look at the changes when you have a moment?

I've done my best to implement the suggested changes, but am sure there's still room for improvement.
A couple of issues I could use your input on:

Some C++ tests are failing in hdfs_test, with the following error:

/arrow/cpp/src/arrow/filesystem/hdfs_test.cc:90: HadoopFileSystem::Make failed, it is possible when we don't have proper 
driver on this node, err msg is IOError: Unable to load libjvm

I'm not sure how best to resolve this — any guidance would be appreciated.

The MSVC compiler is complaining about a forward-declared friend function I'm using in hdfs.cc. Do you have any advice on how to better organise this?

The Python and MATLAB test failures are not related.
Thanks in advance!

pitrou · 2025-05-19T12:38:53Z

Hi @AlenkaF

Some C++ tests are failing in hdfs_test, with the following error:

I think you're misreading the output, the test is actually skipped when the driver fails unloading, which is normal:

(...)
dlopen(/usr/java/latest//lib/amd64/server/libjvm.so) failed: /usr/java/latest//lib/amd64/server/libjvm.so: cannot open shared object file: No such file or directory
/arrow/cpp/src/arrow/filesystem/hdfs_internal.cc:294  try_dlopen(libjvm_potential_paths, "libjvm")
/arrow/cpp/src/arrow/filesystem/hdfs.cc:95  ConnectLibHdfs(&driver_)
/arrow/cpp/src/arrow/filesystem/hdfs.cc:725  ptr->impl_->Init()
/arrow/cpp/src/arrow/filesystem/hdfs_test.cc:205: Skipped
Driver not loaded, skipping

https://github.com/apache/arrow/actions/runs/15109276550/job/42464862030?pr=45998#step:7:3277

The problem is in the other tests, because it seems a destructor crashes:


[ RUN      ] TestHadoopFileSystem.DeleteDirContents
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Running '/build/cpp/debug/arrow-hdfs-test' produced core dump at '/tmp/core.arrow-hdfs-test.25630', printing backtrace:
[New LWP 25630]
[New LWP 25631]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
Core was generated by `/build/cpp/debug/arrow-hdfs-test'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0  0x00007fc37b23d5be in arrow::io::internal::LibHdfsShim::Disconnect () at /arrow/cpp/src/arrow/filesystem/hdfs_internal.cc:340
340	int LibHdfsShim::Disconnect(hdfsFS fs) { return this->hdfsDisconnect(fs); }
[Current thread is 1 (Thread 0x7fc374633dc0 (LWP 25630))]
(...)

https://github.com/apache/arrow/actions/runs/15109276550/job/42464862030?pr=45998#step:7:3281

pitrou · 2025-05-19T12:41:21Z

The MSVC compiler is complaining about a forward-declared friend function I'm using in hdfs.cc. Do you have any advice on how to better organise this?

Hmm, rather than trying to find the exact explanation, a simple solution would be to change these functions into static methods, for example this:

ARROW_EXPORT Status MakeReadableFile(const std::string& path, int32_t buffer_size,
  const io::IOContext& io_context, LibHdfsShim* driver,
  hdfsFS fs, hdfsFile file,
  std::shared_ptr<HdfsReadableFile>* out);

would become:

class ARROW_EXPORT HdfsReadableFile : public RandomAccessFile {
 public:
   (...)

  static Result<std::shared_ptr<HdfsReadableFile>> Make(
      const std::string& path, int32_t buffer_size,
      const io::IOContext& io_context, LibHdfsShim* driver,
      hdfsFS fs, hdfsFile file);

AlenkaF · 2025-05-19T12:41:24Z

Aha, I see! Thanks, will look into it.

AlenkaF · 2025-05-20T13:57:00Z

@pitrou I cleaned up the CI failures (others are not related) and am hoping this changes will not be too bad to review :)

benibus

Thanks! This looks pretty good to me. Just a few comments.

AlenkaF · 2025-06-10T11:53:40Z

@benibus @pitrou would you mind taking another look?

AlenkaF · 2025-06-23T13:44:09Z

@pitrou gentle ping. Would I be too optimistic to try to get it into 21.0.0?

AlenkaF · 2026-09-17T09:58:02Z

@github-actions crossbow submit test-hdfs

github-actions · 2026-09-17T09:58:57Z

Unable to match any tasks for `test-hdfs`
The Archery job run can be found at: https://github.com/apache/arrow/actions/runs/35207971116

Copilot

🔵 Needs a closer look

The refactor introduces a deletion-path bug (URI vs in-filesystem paths) and a couple of build/API issues that should be corrected before merge.

Review details

Suppressed comments (3)

Previously missed (2) — in code that hasn't changed since the last review.

cpp/src/arrow/filesystem/hdfs_internal.cc:39

hdfs_internal.cc uses std::min and std::numeric_limits (in the newly added HdfsReadableFile/HdfsOutputStream code) but doesn't include or . This relies on transitive includes and can break builds as headers change.
cpp/src/arrow/filesystem/hdfs.cc:340
GetChildren() returns the raw HDFS entry name (an absolute "hdfs://..." URI per StatSelector's comment), but DeleteDirContents() later passes those strings to Delete(). This contradicts the earlier guard that in-filesystem APIs must not be passed URIs and can lead to incorrect behavior when deleting directory contents.

cpp/src/arrow/filesystem/hdfs.h:58

HdfsOptions::extra_conf() returns the unordered_map by value, which forces an extra copy on each call (including in HadoopFileSystem::Impl::Init and in the Python bindings). Returning a const reference avoids the copy but would also require updating the Cython declaration in python/pyarrow/includes/libarrow_fs.pxd.

  const std::string& host() const { return host_; }
  int port() const { return port_; }
  const std::string& user() const { return user_; }
  const std::string& kerb_ticket() const { return kerb_ticket_; }
  std::unordered_map<std::string, std::string> extra_conf() const { return extra_conf_; }

Files reviewed: 19/19 changed files
Comments generated: 0 new
Review effort level: Lite

AlenkaF · 2026-09-17T10:00:35Z

@github-actions crossbow submit test-hdfs

github-actions · 2026-09-17T10:02:49Z

Revision: 0625e9e

Submitted crossbow builds: ursacomputing/crossbow @ actions-0d69c54052

Task	Status
test-conda-python-3.11-hdfs-2.10.2
test-conda-python-3.11-hdfs-3.5.0

…he extra_conf situation

AlenkaF · 2026-09-17T11:41:28Z

I added a fix for the 3rd item in last comment from Copilot, connected to extra_conf(), here: 2741e3f

(1st and 2nd items seem like they are dealing with an existing behavior and can be fixed as a follow-up in a separate PR)

AlenkaF · 2026-09-17T11:41:50Z

@github-actions crossbow submit test-hdfs

github-actions · 2026-09-17T11:44:06Z

Revision: 2741e3f

Submitted crossbow builds: ursacomputing/crossbow @ actions-aab95289e3

Task	Status
test-conda-python-3.11-hdfs-2.10.2
test-conda-python-3.11-hdfs-3.5.0

Copilot

🟡 Changes recommended

There are build-breaking issues in the updated bindings/implementation (Cython signature typo and missing standard headers for newly moved C++ code).

Get a fresh assessment by requesting another Copilot review.

Review details

Suppressed comments (1)

cpp/src/arrow/filesystem/hdfs_internal.cc:34

hdfs_internal.cc now contains code using std::numeric_limits and std::min (in the moved HdfsReadableFile/HdfsOutputStream implementations), but it doesn't include <limits> / <algorithm>. This can break compilation depending on transitive includes.

Files reviewed: 19/19 changed files
Comments generated: 1
Review effort level: Lite

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Copilot

🔵 Needs a closer look

There is a compile-breaking namespace/lookup issue in cpp/src/arrow/filesystem/hdfs_internal.cc (and related header hygiene) that must be fixed before merging.

Review details

Suppressed comments (2)

Previously missed (1) — in code that hasn't changed since the last review.

cpp/src/arrow/filesystem/hdfs_internal.cc:37

This file uses std::min / std::numeric_limits in the newly added HDFS file implementations but doesn't include <algorithm> / <limits> directly. Relying on transitive includes is fragile and can break under different standard library/IWYU configurations; add the missing standard headers here.

cpp/src/arrow/filesystem/hdfs_internal.cc:674

GetPathInfoFailed is defined in an anonymous namespace inside arrow::fs::internal, so it isn't reachable as internal::GetPathInfoFailed from namespace arrow::fs. This should fail to compile (no such symbol). Inline the IOErrorFromErrno call here (or move the helper into arrow::fs::internal with external linkage).

Files reviewed: 19/19 changed files
Comments generated: 0 new
Review effort level: Lite

AlenkaF · 2026-09-17T13:59:26Z

OK, this is ready. CI and test-conda-python-3.11-hdfs-* are green.
Last Copilot's comment is:

dealing with unrelated changes (can do a follow-up)
wrong: GetPathInfoFailed is used only in the files where it is defined. Also, the code compiled so no mistake here.

raulcd

I think we should merge this. This is a great simplification and improvement.
Thanks @AlenkaF for following this up!

github-actions Bot added Component: C++ awaiting review Awaiting review Component: Python labels Apr 1, 2025

AlenkaF force-pushed the gh-45747-remove-deprecated-ObjectType-FileStatistics branch from e95f3f3 to 287cb9b Compare April 8, 2025 13:24

AlenkaF changed the title ~~GH-45747: [C++] Remove deprecated ObjectType and FileStatistics~~ GH-45747: [C++] Remove deprecated ObjectType and FileStatistics, refactor hdfs code Apr 9, 2025

AlenkaF marked this pull request as ready for review April 9, 2025 11:40

AlenkaF force-pushed the gh-45747-remove-deprecated-ObjectType-FileStatistics branch from 1bd3f43 to 8e26730 Compare April 28, 2025 05:10

AlenkaF requested review from raulcd and rok as code owners April 28, 2025 05:10

pitrou reviewed Apr 29, 2025

View reviewed changes

github-actions Bot added awaiting committer review Awaiting committer review and removed awaiting review Awaiting review labels Apr 29, 2025

AlenkaF marked this pull request as draft May 19, 2025 07:36

AlenkaF marked this pull request as ready for review May 27, 2025 09:37

AlenkaF requested a review from pitrou May 27, 2025 09:37

AlenkaF force-pushed the gh-45747-remove-deprecated-ObjectType-FileStatistics branch from 72eae6e to 7810940 Compare June 5, 2025 08:38

benibus reviewed Jun 10, 2025

View reviewed changes

Comment thread cpp/src/arrow/filesystem/hdfs_internal.h Outdated

Comment thread cpp/src/arrow/filesystem/hdfs.h Outdated

AlenkaF commented Jun 10, 2025

View reviewed changes

Comment thread python/pyarrow/includes/libarrow_fs.pxd Outdated

AlenkaF force-pushed the gh-45747-remove-deprecated-ObjectType-FileStatistics branch from c7beefc to 281b51a Compare June 23, 2025 13:13

Fix integration, dlopen linkage and mason issues

0625e9e

Copilot AI review requested due to automatic review settings September 17, 2026 09:53

AlenkaF requested review from assignUser, jonkeane and kou as code owners September 17, 2026 09:53

Copilot started reviewing on behalf of AlenkaF September 17, 2026 09:54 View session

Copilot AI reviewed Sep 17, 2026

View reviewed changes

Remove another instance of io hdfs test in integration and clean up t…

2741e3f

…he extra_conf situation

Copilot AI review requested due to automatic review settings September 17, 2026 11:40

Copilot started reviewing on behalf of AlenkaF September 17, 2026 11:41 View session

Copilot AI reviewed Sep 17, 2026

View reviewed changes

Comment thread python/pyarrow/includes/libarrow_fs.pxd Outdated

Fix formatting of extra_conf method declaration

8d861e4

Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>

Copilot AI review requested due to automatic review settings September 17, 2026 13:18

Copilot started reviewing on behalf of AlenkaF September 17, 2026 13:19 View session

Copilot AI reviewed Sep 17, 2026

View reviewed changes

raulcd approved these changes Sep 28, 2026

View reviewed changes

github-actions Bot added awaiting merge Awaiting merge and removed awaiting change review Awaiting change review labels Sep 28, 2026

raulcd merged commit 1369f7f into apache:main Sep 29, 2026
60 of 61 checks passed

raulcd removed the awaiting merge Awaiting merge label Sep 29, 2026

raulcd mentioned this pull request Sep 29, 2026

[C++] Remove deprecated ObjectType and FileStatistics #45747

Closed

AlenkaF deleted the gh-45747-remove-deprecated-ObjectType-FileStatistics branch September 29, 2026 13:20

	if(ARROW_HDFS)
	foreach(ARROW_FILESYSTEM_TARGET ${ARROW_FILESYSTEM_TARGETS})
	target_link_libraries(${ARROW_FILESYSTEM_TARGET} PRIVATE arrow::hadoop)
	endforeach()
	endif()

Conversation

AlenkaF commented Apr 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Rationale for this change

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?

Uh oh!

AlenkaF commented Apr 1, 2025

Uh oh!

pitrou commented Apr 2, 2025

Uh oh!

AlenkaF commented Apr 2, 2025

Uh oh!

AlenkaF commented Apr 9, 2025

Uh oh!

pitrou left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

AlenkaF commented May 19, 2025

Uh oh!

pitrou commented May 19, 2025

Uh oh!

pitrou commented May 19, 2025

Uh oh!

AlenkaF commented May 19, 2025

Uh oh!

AlenkaF commented May 20, 2025

Uh oh!

benibus left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

AlenkaF commented Jun 10, 2025

Uh oh!

AlenkaF commented Jun 23, 2025

Uh oh!

AlenkaF commented Sep 17, 2026

Uh oh!

github-actions Bot commented Sep 17, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🔵 Needs a closer look

Suppressed comments (3)

Uh oh!

AlenkaF commented Sep 17, 2026

Uh oh!

github-actions Bot commented Sep 17, 2026

Uh oh!

AlenkaF commented Sep 17, 2026

Uh oh!

AlenkaF commented Sep 17, 2026

Uh oh!

github-actions Bot commented Sep 17, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟡 Changes recommended

AlenkaF commented Apr 1, 2025 •

edited

Loading