Update RecursiveArrayTools.jl compatibility to version 3

[huiyuxie]

Supplementary fix for #2150 and JoshuaLampert/DispersiveShallowWater.jl#163 (merged).

[github-actions]

Review checklist

This checklist is meant to assist creators of PRs (to let them know what reviewers will typically look for) and reviewers (to guide them in a structured review process). Items do not need to be checked explicitly for a PR to be eligible for merging.

Purpose and scope

• The PR has a single goal that is clear from the PR title and/or description.
• All code changes represent a single set of modifications that logically belong together.
• No more than 500 lines of code are changed or there is no obvious way to split the PR into multiple PRs.

Code quality

• The code can be understood easily.
• Newly introduced names for variables etc. are self-descriptive and consistent with existing naming conventions.
• There are no redundancies that can be removed by simple modularization/refactoring.
• There are no leftover debug statements or commented code sections.
• The code adheres to our conventions and style guide, and to the Julia guidelines.

Documentation

• New functions and types are documented with a docstring or top-level comment.
• Relevant publications are referenced in docstrings (see example for formatting).
• Inline comments are used to document longer or unusual code sections.
• Comments describe intent ("why?") and not just functionality ("what?").
• If the PR introduces a significant change or new feature, it is documented in NEWS.md with its PR number.

Testing

• The PR passes all tests.
• New or modified lines of code are covered by tests.
• New or modified tests run in less then 10 seconds.

Performance

• There are no type instabilities or memory allocations in performance-critical parts.
• If the PR intent is to improve performance, before/after time measurements are posted in the PR.

Verification

• The correctness of the code was verified using appropriate tests.
• If new equations/methods are added, a convergence test has been run and the results
  are posted in the PR.

Created with ❤️ by the Trixi.jl community.

[huiyuxie]

I make the initialization algorithm for DiscreteCallback default to nothing - if that is not what you intend to change with the new struct, this PR definitely won't help.

[huiyuxie]

This will make the current package compatible with Julia >= 1.10 - the CI tests relates to Julia < 1.10 will definitely fail.

[huiyuxie]

The dependency management is really a mess here - for example, the configurations for test env, docs env, and main project env never align. This definitely makes the debug for env configuration hard

[huiyuxie]

Review @ranocha @sloede

[sloede]

This will make the current package compatible with Julia >= 1.10

I don't think we are ready to do this just yet - or did we make a decision about this that I forgot about, @ranocha?

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 84.74%. Comparing base (1488a74) to head (0292601).

❗ There is a different number of reports uploaded between BASE (1488a74) and HEAD (0292601). Click for more details.

HEAD has 5 uploads less than BASE

Flag	BASE (1488a74)	HEAD (0292601)
unittests	26	21

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #2194       +/-   ##
===========================================
- Coverage   96.40%   84.74%   -11.66%     
===========================================
  Files         485      485               
  Lines       38918    38874       -44     
===========================================
- Hits        37517    32941     -4576     
- Misses       1401     5933     +4532     

Flag	Coverage Δ
unittests	84.74% <ø> (-11.66%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[huiyuxie]

Then the issue #1789 will be delayed - when do you plan to upgrade to Julia 1.10?

[sloede]

Then the issue #1789 will be delayed - when do you plan to upgrade to Julia 1.10?

Which version of RecursiveArrayTools.jl is really required? Just v3 (i.e., v3.0.0 would be sufficient) or a specific, later version? With, e.g., v3.2 we would still be able to keep Julia v1.9 compatibility.

[huiyuxie]

Good question 👍let me check

[huiyuxie]

Which version of RecursiveArrayTools.jl is really required?

I have no idea, I just know that >= 2.38.10 does not work and >= 3 works. Can you answer Michael's question @jlchan - I don't know which version you actually need

[ranocha]

I would hesitate to require Julia 1.10 and newer. We have seen some performance impacts in HPC settings, e.g., JuliaLang/julia#55009, JuliaLang/julia#50985

[jlchan]

Which version of RecursiveArrayTools.jl is really required?

I have no idea, I just know that >= 2.38.10 does not work and >= 3 works. Can you answer Michael's question @jlchan - I don't know which version you actually need

We'd need at least RecursiveArrayTools.jl 3.27.1, which is where @huiyuxie's PR fixing VectorOfArray was implemented.

[sloede]

Which version of RecursiveArrayTools.jl is really required?

I have no idea, I just know that >= 2.38.10 does not work and >= 3 works. Can you answer Michael's question @jlchan - I don't know which version you actually need

We'd need at least RecursiveArrayTools.jl 3.27.1, which is where @huiyuxie's PR fixing VectorOfArray was implemented.

Too bad. This is already at Julia v1.10 😞

[huiyuxie]

I would hesitate to require Julia 1.10 and newer. We have seen some performance impacts in HPC settings, e.g., JuliaLang/julia#55009, JuliaLang/julia#50985

Does it mean you prefer to upgrade to version 1.10 until both of these issues are resolved?

[ranocha]

That's something we need to discuss.

[JoshuaLampert]

This is a bad situation. IMHO, we cannot wait until these two issues are resolved before we fix the incompatibility with newer versions of the whole SciML stack. Fixing this feels more and more urgent and it also looks like the two julia issues will not be fixed anytime soon.
Is there any chance to keep the compat bound as v1.8 for julia, but also allow old and new versions of RecursiveArrayTools.jl (something like `RecursiveArrayTools = "2.38.10, 3")? Then one would need to implement different behavior depending on the version of RecursiveArrayTools.jl, i.e. for RecursiveArrayTools.jl < v3.27.1 we use the current version and for RecursiveArrayTools.jl >= v3.27.1 we implement the new version. In that case for julia < v1.10 the old version would be used and for julia >= v1.10 the new one. However, if that is really practically viable is another question. Do you see any major problem that rules this solution out?

[huiyuxie]

we would need to figure out a way to determine the version of RecusiveArrayTools.jl at compile time.

FIX: at precompile time

[ranocha]

Thanks! Is the following summary correct?

• This PR (updating RecursiveArrayTools.jl) is needed to continue working on Use VectorOfArray in wrap_array for DGMulti solvers #2150. To do so, we need to drop support of Julia versions older than 1.10.
• Use VectorOfArray in wrap_array for DGMulti solvers #2150 is needed to fix issues such as Workaround for SciML dropping support of arrays of SVectors #1789 and remove the hotfix

  Trixi.jl/Project.toml

  Line 12 in ab8ecad

  DiffEqBase = "2b5f629d-d688-5b77-993f-72d75c75574e"

  Trixi.jl/Project.toml

  Line 73 in ab8ecad

  DiffEqBase = "6 - 6.143"

  Trixi.jl/.github/workflows/Downgrade.yml

  Line 75 in ab8ecad

  skip: LinearAlgebra,Printf,SparseArrays,UUIDs,DiffEqBase,DelimitedFiles

• These steps are required to use new versions of the SciML packages, e.g., the low-dependency packages for the time integration methods we are typically using

[JoshuaLampert]

Thanks! Is the following summary correct?

Yes, that's also how I understand it.

[huiyuxie]

e.g., the low-dependency packages for the time integration methods we are typically using

Which one are you referring to?

And how do you display PRs and issues with graphics and titles instead of just numbers?

@ranocha

[ranocha]

e.g., the low-dependency packages for the time integration methods we are typically using

Which one are you referring to?

OrdinaryDiffEqLowStorageRK.jl and OrdinaryDiffEqSSPRK.jl

And how do you display PRs and issues with graphics and titles instead of just numbers?

Thanks! Is the following summary correct?

- This PR (updating RecursiveArrayTools.jl) is needed to continue working on https://github.com/trixi-framework/Trixi.jl/pull/2150. To do so, we need to drop support of Julia versions older than 1.10.
- https://github.com/trixi-framework/Trixi.jl/pull/2150 is needed to fix issues such as https://github.com/trixi-framework/Trixi.jl/issues/1789 and remove the hotfix https://github.com/trixi-framework/Trixi.jl/blob/ab8ecad4249fa718fab0ba6acba60bc5d92b8a75/Project.toml#L12 https://github.com/trixi-framework/Trixi.jl/blob/ab8ecad4249fa718fab0ba6acba60bc5d92b8a75/Project.toml#L73 https://github.com/trixi-framework/Trixi.jl/blob/ab8ecad4249fa718fab0ba6acba60bc5d92b8a75/.github/workflows/Downgrade.yml#L75
- These steps are required to use new versions of the SciML packages, e.g., the low-dependency packages for the time integration methods we are typically using

[huiyuxie]

Interesting 🤔

[huiyuxie]

So when do you @ranocha plan to upgrade to Julia 1.10? I need latest version of CUDA.jl to gain higher performance.

[sloede]

I'm wondering what would happen if we just add v3 to the compat list - wouldn't that allow us to still keep using Julia v1.9 (but with the older SciML stack) but also v1.10/11 with the newer stack?

Or are there other modifications necessary in the code to support the new code? If yes - are they everywhere or limited to certain features. That is, would be get away by declaring, e.g., that some parts of DGMulti no longer run with Julia v1.9?

[jlchan]

Or are there other modifications necessary in the code to support the new code? If yes - are they everywhere or limited to certain features. That is, would be get away by declaring, e.g., that some parts of DGMulti no longer run with Julia v1.9?

I think it would be complicated. All parts of DGMulti would need to be marked unless compat was set very carefully, and even if the compat worked correctly, there would still need to be some branching code depending on the Julia version.

[sloede]

Or are there other modifications necessary in the code to support the new code? If yes - are they everywhere or limited to certain features. That is, would be get away by declaring, e.g., that some parts of DGMulti no longer run with Julia v1.9?

I think it would be complicated. All parts of DGMulti would need to be marked unless compat was set very carefully, and even if the compat worked correctly, there would still need to be some branching code depending on the Julia version.

OK. Without having really decided for myself whether I think it's acceptable from a maintenance POV: Would you be OK with making DGMulti effectively work only with Julia v1.10+ or is that something you would like to avoid for now?

[jlchan]

I'd be perfectly fine with that! I don't think there are a lot (or really any?) current users, so I'm happy to restrict it to 1.10+.

[ranocha]

So the suggestion is to allow Julia 1.9 but DGMulti only on Julia 1.10? Would this work? As far as I understand, we need a new RecursiveArrayTools.jl version for it that supports only Julia 1.10. Thus, if DGMulti is part of Trixi.jl, we need to depend on a version of RecusiveArrayTools.jl that requires Julia 1.10.

Since Julia 1.9 is not supported by bugfixes etc. anymore, I tend to prefer upgrading to Julia 1.10. People who need/want to use Julia 1.9 can still continue to use older versions of Trixi.jl. But that's something at least @trixi-framework/principal-developers should agree with.

[huiyuxie]

I'm wondering what would happen if we just add v3 to the compat list - wouldn't that allow us to still keep using Julia v1.9 (but with the older SciML stack) but also v1.10/11 with the newer stack?

RecursiveArrayTools.jl v3 requires a higher version of SciMLBase.jl to precompile without errors. I tried using SciMLBase.jl v2.67.0, which was set by @jlchan in #2150. However, this version already requires Julia v1.10 to be compatible, see https://github.com/SciML/SciMLBase.jl/blob/v2.67.0/Project.toml.

Maybe a version lower than v2.67.0 works good with RecursiveArrayTools.jl v3 and also is compatible with Julia v1.9 but as I mentioned here #2197 - this makes narrowing down the exact version quite time-consuming.

[huiyuxie]

Or are there other modifications necessary in the code to support the new code?

Makie.jl has to be updated to v0.21 to adapt SciMLBase.jl v2.67.0 (and also CairoMakie.jl) - some parts of code need change.

[jlchan]

Thanks for responding @huiyuxie; sorry I haven't made more progress on my end yet, I've been sick since last week.

Edit: wait, this isn't my PR. Sorry, I am still kind of out of it

[huiyuxie]

Take a rest first - it was just a mention, not asking for a response.

[ranocha]

We have discussed this issue with some main developers and agreed that we will go ahead with updating to Julia version 1.10.

[ranocha]

Thanks a lot! We have to update the CI setup to remove tests on older versions of Julia. We should also mention the update in the NEWS.md.

[huiyuxie]

Sure

[huiyuxie]

Do you @ranocha prefer to completely delete legacy tests or just simply remove them from the CI?

[sloede]

Do you @ranocha prefer to completely delete legacy tests or just simply remove them from the CI?

We'll do the bump to v1.10 in a separate PR first, see #2214. Then we can proceed with this PR without having to deal with multiple unrelated issues simultaneously.