ARROW-11387: [Rust] fix build for conditional compilation of features 'simd + avx512'

[ritchie46]

This fixes conditional compilation for simd and avx512. Currently compilation for both fails:

error[E0428]: the name `buffer_bin_and` is defined multiple times
--> arrow/src/buffer.rs:472:1
|
414 | / pub(super) fn buffer_bin_and(
415 | | left: &Buffer,
416 | | left_offset_in_bits: usize,
417 | | right: &Buffer,
... |
468 | | }
469 | | }
| |_- previous definition of the value `buffer_bin_and` here
...
472 | / pub(super) fn buffer_bin_and(
473 | | left: &Buffer,
474 | | left_offset_in_bits: usize,
475 | | right: &Buffer,
... |
501 | | }
502 | | }
| |_^ `buffer_bin_and` redefined here
|
= note: `buffer_bin_and` must be defined only once in the value namespace of this module
error[E0428]: the name `buffer_bin_or` is defined multiple times
--> arrow/src/buffer.rs:583:1
|
525 | / pub(super) fn buffer_bin_or(
526 | | left: &Buffer,
527 | | left_offset_in_bits: usize,
528 | | right: &Buffer,
... |
579 | | }
580 | | }
| |_- previous definition of the value `buffer_bin_or` here
...
583 | / pub(super) fn buffer_bin_or(
584 | | left: &Buffer,
585 | | left_offset_in_bits: usize,
586 | | right: &Buffer,
... |
612 | | }
613 | | }
| |_^ `buffer_bin_or` redefined here
|
= note: `buffer_bin_or` must be defined only once in the value namespace of this module

This PR is tested on:

$ cargo c --features "simd"

$ cargo c --features "simd avx512"

$ cargo c --features "avx512"

Maybe a check like used in cargo hack can be added to the CI pipeline? This checks compilation for multiple feature combinations.

[github-actions]

Thanks for opening a pull request!

Could you open an issue for this pull request on JIRA?
https://issues.apache.org/jira/browse/ARROW

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

ARROW-${JIRA_ID}: [${COMPONENT}] ${SUMMARY}

See also:

• Other pull requests
• Contribution Guidelines - How to contribute patches

[codecov-io]

Codecov Report

Merging #9337 (68c60cc) into master (ab5fc97) will not change coverage.
The diff coverage is n/a.

@@ Coverage Diff @@## master #9337 +/- ##
=======================================
Coverage 81.89% 81.89% =======================================
Files 215 215 Lines 52988 52988 =======================================
Hits 43392 43392 Misses 9596 9596 

Impacted Files	Coverage Δ
rust/arrow/src/buffer.rs	96.21% <ø> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ab5fc97...68c60cc. Read the comment docs.

[ChristianBeilschmidt]

Thanks for opening a pull request!

Could you open an issue for this pull request on JIRA?
https://issues.apache.org/jira/browse/ARROW

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

ARROW-${JIRA_ID}: [${COMPONENT}] ${SUMMARY}

See also:

* [Other pull requests](https://github.com/apache/arrow/pulls/)
* [Contribution Guidelines - How to contribute patches](https://arrow.apache.org/docs/developers/contributing.html#how-to-contribute-patches)

@ritchie46 I guess you need to prepend your Jira Issue to your PR-title:
https://issues.apache.org/jira/browse/ARROW-11387

So, I guess
ARROW-11387: [Rust] fix build for conditional compilation of features 'simd + avx512'

And thank you for the PR!

[alamb]

It looks good to me. Thanks @ritchie46

FYI @vertexclique

[alamb]

On my machine, specifically, I tested in this manner:

# works fine
cargo +nightly build --features=simd
# works fine
cargo +nightly build --features=avx512
# fails without this PR, works with this PR:
cargo +nightly build --features=avx512,simd

[vertexclique]

I think we've talked about this mutual exclusivity of features here: #8665 (comment)
I think this pr settles down simd feature's dependencies when avx512 is used. When this PR is merged it would be nice to document which one got selected. Like:

cargo +nightly build --features=avx512,simd This selects avx512 with this pr in I presume. That's enough to document from my point of view.

[github-actions]

https://issues.apache.org/jira/browse/ARROW-11387

[ChristianBeilschmidt]

Will this be published as a patch version to 3.0.0?

[alamb]

Will this be published as a patch version to 3.0.0?

@ChristianBeilschmidt -- I do not know what the plans are for a 3.0.0 patch. @andygrove do you know of how/if we do such a thing?

[andygrove]

For any JIRA issues that we want to include in a patch release, we should update the release version to include both 3.0.0 and 3.0.1 so that we have the list of candidates to use if we create a patch release as the first step

[jorgecarleitao]

I assigned both 4.0.0 and 3.0.1 to it, then. I agree that is a fix for a patch.