Cast max_stepsize in Mesh Adaptive Direct Seach (#526)

[sblelong]

• Explicit cast of MADS' maximum stepsize to match the mesh scaling parameter type
• Unit test of MADS running on a sphere

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.78%. Comparing base (837bb31) to head (8905535).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #527   +/-   ##
=======================================
  Coverage   99.78%   99.78%           
=======================================
  Files          86       86           
  Lines        9781     9781           
=======================================
  Hits         9760     9760           
  Misses         21       21           

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[kellertuer]

Thanks, I did not know the function oftype! I have just one comment on the code PR itself.

For the failing tests

• please add a small entry to the Changelog.md
• make sure to run the runic code formatter on the code. “Usually” it would report errors here, but for external contributions, that is not the case here, but in the test itself you can also see what went wrong, see https://github.com/JuliaManifolds/Manopt.jl/actions/runs/18887296259/job/53905822587#step:3:201 – you just missed spacing at one keyword it seems :)

[sblelong]

Thanks a lot for your quick feedback! I was actually trying to figure out how to reach out to you on Slack or Zulip since the failing tests led to some panicking on my side... It is my first PR on a fairly large repo so, thank you for bearing with me. :)

I'm working on fixing those two errors.

I may have missed something, but I think the contributing guidelines ask to run JuliaFormatter.jl instead of Runic.jl. I guess the switch to runic was quite recent, maybe not yet reverberated to the guidelines?

[kellertuer]

Oh, did I miss to update the guidelines, sorry for that! Will check that tomorrow – and fix that if necessary;
I also am already on mobile and have turned of slack and Zulip (in CET its 9:30pm).

I'm sure you'll manage, no reason to panic - just keep calm and code on :)

[sblelong]

There's absolutely no harm in that, the fact that all files of the repo were modified when applying JuliaFormatter should've been a hint for me that another formatter was in use...

Thanks all the more so that it's getting late! Hope those failing tests weren't too much of a disturbance in the evening. I'll take advantage of the remaining sun time I have to fix those issues.

[kellertuer]

Oh usually the PR fixing an issue would close the issue, now you closed it a bit early, but never mind :) (if you mention in the first comment or somewhen along that this PR closes/fixes #anothernumber then that automatically closes it)

[sblelong]

Oops... Sorry again. I feared keeping it open would be confusing since the issue is somehow addressed. But that's one more good practice I'll keep in mind for the future.

[kellertuer]

It is not wrong, just a bit unusual, for example if this would only fix something half ;)

[kellertuer]

I did adapt my proposal of the tests above and the tests now also directly use Manifolds.jl 0.11.2, but we could enforce that a bit better by changing the compat entry in Project.toml as well.

[sblelong]

Indeed, using Manifolds.jl 0.11.2 solves that. The new tests seem to pass perfectly.

[kellertuer]

Thanks again. I will include this in the next release :)

[sblelong]

My pleasure :)