Add .singleOptional() to Mono and Flux

[AndreasHuber-CH]

Fixes #3311

Signed-off-by: Andreas Huber andi@hubervonberg.ch

[AndreasHuber-CH]

I've started off by duplicating the existing code for the single() use case and adapting it for the optional use case.
This allowed me also to copy over most unit tests.
But by doing that I ended up supporting Flux.singleOptional() use case as well.

If you prefer not including the Flux part, I can delete the Flux.singleOptional(), move corresponding code from the MonoSingleOptional to MonoSingleOptionalMono, and the same for unit tests.
Would you prefer that?

[AndreasHuber-CH]

Also, I haven't started designing marbles yet that stand for Optional.of() and Optional.empty(). Or could someone help me with that?

[chemicL]

@AndreasHuber-CH coming back after a break - thanks for the PR! Unless you have a justified use case for Flux.singleOptional(), please remove the implementation. In my head, Flux.collectList() is the logical counterpart for a sequence of items. If it's needed, we can add it in the future.
For the marble diagrams, I haven't created any yet so I'll get back to you once I clarify what the process is.
I'll go ahead and do a review once you remove the Flux implementations and we can follow up with the marbles after the initial review round.

[chemicL]

For marble diagrams, it occurs the team has been using Inkscape and more detailed instructions are in https://github.com/reactor/reactor-core/tree/main/docs/svg - the conventions.svg file can be used as the source of visual components for creating a new marble diagram.

[AndreasHuber-CH]

I removed the Flux.singleOptional() use case. The code is now ready to be reviewed.
For the svg, I'll try it out, but don't have a proposition yet.

[chemicL]

Thank you for the changes. I did review the code and left a few comments to address :) Good luck with the marble diagram, hope it works out!

[chemicL]

Also, please do validate the build result and run spotlessApply gradle job.

[AndreasHuber-CH]

sorry about my last commit. It added useless whitespaces changes.
I imported the file https://github.com/reactor/reactor-core/blob/main/codequality/eclipse/reactor.style.xml into eclipse as I saw I was using "spaces" vs "tabs" used in other files. But it seems that messed up more than it fixed. Though in my local I didn't spot the newline changes... I'll revert

[AndreasHuber-CH]

OK I fixed the windows line ending issue. Actually it wasn't the eclipse formatter, but when "gradlew spotlessApply" is run on Windows, it changes line endings of all files it updates from Linux to Windows line endings.
(And I tried running it from WSL, but due to slower file access the command was taking forever)

[chemicL]

OK I fixed the windows line ending issue. Actually it wasn't the eclipse formatter, but when "gradlew spotlessApply" is run on Windows, it changes line endings of all files it updates from Linux to Windows line endings.
(And I tried running it from WSL, but due to slower file access the command was taking forever)

Not sure whether that can help (I'm not a Windows user), but perhaps a git configuration could fix such issues in the future:
https://community.openhab.org/t/spotless-converting-newlines-to-crlf/109493

[chemicL]

LGTM!

[chemicL]

I wrote a few JMH benchmarks to validate we indeed have a speedup over a combination of operators:

Benchmark                                          Mode  Cnt   Score   Error  Units
MonoOptionalBenchmark.singleOptional               avgt    5   4.284 ± 0.096  ns/op
MonoOptionalBenchmark.singleOptionalBaseline       avgt    5  38.576 ± 0.091  ns/op
MonoOptionalBenchmark.singleOptionalEmpty          avgt    5   0.591 ± 0.006  ns/op
MonoOptionalBenchmark.singleOptionalEmptyBaseline  avgt    5  32.916 ± 0.343  ns/op
MonoOptionalBenchmark.singleOptionalError          avgt    5  23.348 ± 0.165  ns/op
MonoOptionalBenchmark.singleOptionalErrorBaseline  avgt    5  59.782 ± 0.774  ns/op

Source: https://gist.github.com/chemicL/67b4f87a58950ae6e5d4dbcd4d2564b7

[chemicL]

Thank you for the contribution @AndreasHuber-CH! The PR is merged and will become part of the next release. 🚀