Move codegen into src/ folder

[avantgardnerio]

Which issue does this PR close?

Closes #3332.

Rationale for this change

Described in issue.

What changes are included in this PR?

A change to the generated code location.

Are there any user-facing changes?

no

[andygrove]

I would love to see code completion working with protobuf generated code. That has been frustrating not being able to do that.

The only concern I have with this change is that cargo clean will no longer remove generated source files. I don't know how much of a problem that will be in practice though.

[avantgardnerio]

Example of it working in CLion:

[codecov-commenter]

Codecov Report

Merging #3333 (8ffc83f) into master (b175f9a) will increase coverage by 0.00%.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #3333   +/-   ##
=======================================
  Coverage   85.48%   85.48%           
=======================================
  Files         294      294           
  Lines       54115    54100   -15     
=======================================
- Hits        46259    46248   -11     
+ Misses       7856     7852    -4     

Impacted Files	Coverage Δ
datafusion/core/tests/sql/timestamp.rs	99.65% <ø> (-0.01%)	⬇️
datafusion/proto/src/lib.rs	93.52% <ø> (ø)
datafusion/expr/src/logical_plan/plan.rs	77.77% <0.00%> (+0.33%)	⬆️
datafusion/core/src/physical_plan/metrics/value.rs	87.43% <0.00%> (+0.50%)	⬆️
datafusion/expr/src/window_frame.rs	93.27% <0.00%> (+0.84%)	⬆️

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[avantgardnerio]

@tustvold @alamb What do you two think? This seems (to me) to be the best of all worlds:

1. generated code is not checked in
2. generated code is in source folders so IDEs work again

The only downside is that protoc is required, but that was already a requirement for datafusion. If anyone is interested I can (in another PR), have build.rs download and unzip the appropriate protoc for the platform.

[alamb]

I will review this in more detail, likely tomorrow morning.

…

On Thu, Sep 1, 2022 at 5:11 PM Brent Gardner ***@***.***> wrote: @tustvold <https://github.com/tustvold> @alamb <https://github.com/alamb> What do you two think? This seems (to me) to be the best of all worlds: 1. generated code is not checked in 2. generated code is in source folders so IDEs work again The only downside is that protoc is required, but that was already a requirement for datafusion. If anyone is interested I can (in another PR), have build.rs download and unzip the appropriate protoc for the platform. — Reply to this email directly, view it on GitHub <#3333 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AADXZMOFBEEMKESLDLOXYOTV4ELW3ANCNFSM6AAAAAAQCTKVW4> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[alamb]

Sorry I ran out of time today to give this a careful look. It is on my list...

There was all sorts of good stuff happening in DataFusion I still haven't fully caught up

[alamb]

I like this idea and I think we should try it. If something about generating the code into the source tree causes unforseen issues we can always roll this back.

Thanks a lot @avantgardnerio

[alamb]

This is needed so it works on a clean checkout, right?

[avantgardnerio]

The CI jobs are parallelized, so Lint doesn't get a copy of the files generated by Build, but lint did complain I did a mod datafusion for a file that didn't exist. Since this is a CI-only issue, this seemed like an acceptable fix for CI ... ?

[alamb]

This PR appears to need a merge or rebase against master

[alamb]

I believe these changes were also included in #3311 so closing this PR -- @avantgardnerio please let me know if that is incorrect and reopen