[Java] Move (most of) the name mangling logic to the Refiner

[andreaTP]

Sharing to receive early feedback on the direction (cc. @baywet).

I started with Java(instead of Python where we experienced more issues) as I felt more comfortable with it, but I believe that this effort should be extended to all of the languages.

You can check the results by running:

• on the main branch: dotnet publish ./src/kiota/kiota.csproj -c Release -p:PublishSingleFile=true -p:PublishReadyToRun=true -o ./publish
• on this PR branch I'm testing with commands like: (cd src/kiota && dotnet build) && ./it/compare-generation.ps1 -descriptionUrl "oas::petstore" -language java
• use more complex specs by testing against the rest ...

[andreaTP]

I feel good about the direction and the diff is confirming some improvements, but I haven't verified yet that everything is consistent ...

[baywet]

Thanks for starting this

[andreaTP]

@baywet and @andrueastman this should be ready for review now 🙂 .

[baywet]

@andreaTP a couple of integration tests for Java are failing, can you look into it before we review please?

[andreaTP]

The failures looks legit, I probably missed something in the Error classes, I'll have a look 👍

[andreaTP]

Ready for review now 👍 happy that the IT tests have detected this last detail 🙂 .

for my future self, the incantation to run all the integration tests for one language:

yq -e '.jobs.integration.strategy.matrix.description[]' .github/workflows/integration-tests.yml | xargs -I{} ./it/generate-code.ps1 -descriptionUrl '{}' -language java && ./it/exec-cmd.ps1 -descriptionUrl '{}' -language java

[andreaTP]

ready for the next review round 👍

[baywet]

Thanks for this great contribution!
Will you be doing the same for other languages?
@andrueastman for final review and merge

[andreaTP]

Will you be doing the same for other languages?

I started this with the intention to get some traction on the subject and ... help 🙂
If no one is picking it up, I can try to go through it, but it's going to take time ...
We also need to properly review new developments to avoid regressions (would it be possible to automate? a source Generator can probably do it ...).

[baywet]

We should be able to provide help in validating the changes.
Help in implementing those will be harder, the top priority is to get more languages to a stable maturity level at this point.
I'm curious about the idea of source generators, can you expand?

[andreaTP]

We should be able to provide help in validating the changes.

This is already super valuable help!

Help in implementing those will be harder, the top priority is to get more languages to a stable maturity level at this point.

I would argue that removing a design issue in the implementation is pretty important to be confident about the maturity level, as, in this case, it's excluding a lot of subtle bugs that are hard to find, hard to debug, and, ultimately, hard to solve (workarounds are very likely to introduce regressions).
But this is on you to prioritize 🙂 ❤️

At my own peace, I'll try to catch up with the other languages 🙂

I'm curious about the idea of source generators, can you expand?

Coming from Scala we are amused to solve all of the problems with macros ...
In the docs we have:

A generator can analyze your source code and emit the code it needs to "wire up" your app.

This means that a Source Generator can analyze the source code and look up for usages of i.e. StringExtensions in the Writer package, and either:

• fail loudly
• invalidate the build by emitting invalid code

[baywet]

Did you mean a code analyzer instead by any chance?

[andreaTP]

That's correct, I wasn't aware of code analyzers, the same can likely be done inside a Source Generator, but the analyzer should suffice.

[baywet]

Added a note in #2842

[andrueastman]

Looks great @andreaTP