Expose GNC params to python

[amadoantonini]

No description provided.

[varunagrawal]

@amadoantonini thanks for the PR. Can you please add some python unit tests to verify everything works as expected?

[amadoantonini]

@varunagrawal added some tests, let me know if those are ok.

[varunagrawal]

Awesome stuff @amadoantonini

[varunagrawal]

@amadoantonini can you please resolve the CI failures?

[amadoantonini]

I can try. Any ideas as to why those seemingly unrelated tests fail?

[ProfFan]

I think this is the problem:

Traceback (most recent call last):
[1981](https://github.com/borglab/gtsam/runs/7333219428?check_suite_focus=true#step:7:1982)
  File "/Users/runner/work/gtsam/gtsam/python/gtsam/tests/test_NonlinearOptimizer.py", line 128, in test_gnc_params
[1982](https://github.com/borglab/gtsam/runs/7333219428?check_suite_focus=true#step:7:1983)
    params.setKnownInliers(inl)
[1983](https://github.com/borglab/gtsam/runs/7333219428?check_suite_focus=true#step:7:1984)
TypeError: setKnownInliers(): incompatible function arguments. The following argument types are supported:
[1984](https://github.com/borglab/gtsam/runs/7333219428?check_suite_focus=true#step:7:1985)
    1. (self: gtsam.gtsam.GncLMParams, knownIn: std::__1::vector<unsigned long, std::__1::allocator<unsigned long> >) -> None

std::__1::vector<unsigned long, std::__1::allocator<unsigned long> > need to have a type alias and wrapped. This is also essential for the MATLAB wrapper to work.

[ProfFan]

P.S. I have 0 idea why the automatic binding code cause so many hard-to-debug issues. I encountered an infinity loop issue that is caused by stl.h when wrapping the hybrid code, took me a week to debug. Should we try upgrade the pybind11 lib and see if problems disappear? @varunagrawal

[varunagrawal]

This looks like the Jacobians are not being copied back to the C++ side of things. I guess we have to look at how the opacity is being set.

This is also a good time to revisit my PR on using the stl bindings fully.

[varunagrawal]

@amadoantonini the issue seems to be tangential to your fix. @ProfFan and I are working on figuring this out, so we thank you for your patience!

[varunagrawal]

@amadoantonini I opened a PR to fix your issue openspacelabs#1
Can you please merge it in? It should automatically reflect here.

[varunagrawal]

Ahhh I'm going to have to look into this more.

[amadoantonini]

Thanks @varunagrawal! Let me know if I can help.

[varunagrawal]

@amadoantonini can you please merge in the latest develop? We made some CI changes to fix issues we've been seeing lately (this PR included).

[amadoantonini]

That seems to have fixed it on my side!

[varunagrawal]

@amadoantonini gentle reminder about openspacelabs#2

[varunagrawal]

@amadoantonini can you please merge in the latest develop? Github Actions is being funky.

[amadoantonini]

@varunagrawal I had to resolve a conflict. I think I did the right thing based on the order of your changes. Please verify.

[varunagrawal]

Yup looks good to me.

[varunagrawal]

@amadoantonini please merge this openspacelabs#3 and please allow maintainers to make edits so I can push here directly.

[varunagrawal]

Yay finally merged!

[amadoantonini]

Thanks for all the support @varunagrawal !