MINOR: [C#] Handle Empty Schema

[gloomweaver]

Rationale for this change

While developing SDK encountered this error:

System.ArgumentOutOfRangeException : Specified argument was out of the range of valid values. (Parameter 'length')
   at Apache.Arrow.Flight.FlightMessageSerializer.DecodeSchema(ReadOnlyMemory`1 buffer)

The issue is the schema buffer is empty, and other libraries don't throw the error in this case. This PR makes this consistent with libraries for other languages.

What changes are included in this PR?

Return null if Schema buffer is empty.

Are these changes tested?

Tested against our Flight API.

Are there any user-facing changes?

Fix critical bug this inability to run request against flight service, which doesn't return schema in GetFlightInfo.

[github-actions]

Thanks for opening a pull request!

If this is not a minor PR. Could you open an issue for this pull request on GitHub? https://github.com/apache/arrow/issues/new/choose

Opening GitHub issues ahead of time contributes to the Openness of the Apache Arrow project.

Then could you also rename the pull request title in the following format?

GH-${GITHUB_ISSUE_ID}: [${COMPONENT}] ${SUMMARY}

or

MINOR: [${COMPONENT}] ${SUMMARY}

In the case of PARQUET issues on JIRA the title also supports:

PARQUET-${JIRA_ISSUE_ID}: [${COMPONENT}] ${SUMMARY}

See also:

• Other pull requests
• Contribution Guidelines - Contributing Overview

[gloomweaver]

@CurtHagenlocher When can we expect the next release, this issue is a blocker.

[CurtHagenlocher]

@CurtHagenlocher When can we expect the next release, this issue is a blocker.

Arrow releases roughly every three months. The next release is expected at the beginning of August.

[gloomweaver]

Can you do minor release sooner to include this fix?

[CurtHagenlocher]

Can you do minor release sooner to include this fix?

At this time, there's no way to publish a build of just the C# components. An out-of-band release of Arrow (i.e. a "dot one") is quite a heavy process -- effectively the same work as a full release -- and so it's typically reserved for regressions from a previous release. If it turns out that there is a 16.2, I can propose this as a candidate for inclusion.

[gloomweaver]

Sure, sounds good.

[conbench-apache-arrow]

After merging your PR, Conbench analyzed the 5 benchmarking runs that have been run so far on merge-commit 3333648.

There was 1 benchmark result indicating a performance regression:

• Commit Run on test-mac-arm at 2024-06-14 20:57:53Z
  • tpch (R) with engine=arrow, format=native, language=R, memory_map=False, query_id=TPCH-20, scale_factor=1

The full Conbench report has more details.