Fixes

[varunagrawal]

Fixes #1738 and #1982.

Additionally, replaced [=], which captures by copy in lambda expressions, by [&] which captures by reference.

I updated the CI to check for lower and upper bounds of versions, with the assumption that everything in between is supported. Additionally, if the no_boost flag is set, I updated the CI to not install Boost so that we can better capture issues with Boost being optional.

Update: Also fixed a bug from #1444 where while the $GITHUB_ENV variable is updated for the No-Boost case, it isn't actually used anywhere.

cc: @Gold856

[thorstink]

Maybe nitpicking, by when trying cpp20, this this deprecation warning now also pops up; I think this is the only one as I was able to compile using cpp20 before these constraint-features got merged. Not sure if it makes sense to add a cpp20 pipeline, AFAIK it would compile except for this.

18.23 /tmp/gtsam/gtsam-0959bfef5c7e206a78e38ded84c4397940ea80ac/gtsam/constrained/InequalityPenaltyFunction.cpp: In lambda function:
18.23 /tmp/gtsam/gtsam-0959bfef5c7e206a78e38ded84c4397940ea80ac/gtsam/constrained/InequalityPenaltyFunction.cpp:25:10: error: implicit capture of 'this' via '[=]' is deprecated in C++20 [-Werror=deprecated]
18.23    25 |   return [=](const double& x, OptionalJacobian<1, 1> H = {}) -> double { return (*this)(x, H); };

[dellaert]

Maybe nitpicking, by when trying cpp20, this this deprecation warning now also pops up; I think this is the only one as I was able to compile using cpp20 before these constraint-features got merged. Not sure if it makes sense to add a cpp20 pipeline, AFAIK it would compile except for this.

18.23 /tmp/gtsam/gtsam-0959bfef5c7e206a78e38ded84c4397940ea80ac/gtsam/constrained/InequalityPenaltyFunction.cpp: In lambda function:
18.23 /tmp/gtsam/gtsam-0959bfef5c7e206a78e38ded84c4397940ea80ac/gtsam/constrained/InequalityPenaltyFunction.cpp:25:10: error: implicit capture of 'this' via '[=]' is deprecated in C++20 [-Werror=deprecated]
18.23    25 |   return [=](const double& x, OptionalJacobian<1, 1> H = {}) -> double { return (*this)(x, H); };

@varunagrawal the fix is easy: just replace that line with

return [this](const double& x, OptionalJacobian<1, 1> H = {}) { return this->operator()(x, H); };

BTW, there are some other occurrences of [] in an example and in some tests. In these (few) cases just narrowing the scope (e.g. [=] -> [nearZero] in testSO3.cpp) should resolve this.

[dellaert]

Maybe also merge in develop after #1983
I'll approve after suggested [=] fixes and CI changes, please re-request a review from me then.

[dellaert]

Awesome, just some doubt about removing default copy constructor.

[dellaert]

I think I just reviewed a PR that added these?

[varunagrawal]

This copy constructor was added 2 years ago.

I am currently getting an error because of this

error: definition of implicit copy assignment operator for 'TestOptionalStruct' is deprecated because it has a user-declared copy constructor [-Werror,-Wdeprecated-copy]

Seems to be that using the compiler generated copy-constructors and using the = default constructor no longer gel well together.

[dellaert]

Seems like not installing BOOST already found an issue with our cmake. Happy for this PR to omit that change and surface it in an issue.

[varunagrawal]

Seems like not installing BOOST already found an issue with our cmake. Happy for this PR to omit that change and surface it in an issue.

Ah I was a bit too quick for you. Already fixed it here.