Moving to `std::shared_ptr`

[humanoid-path-planner]

Moving to `std::shared_ptr`

[jcarpent]

@nim65s You can now move on to this issue. #260 forces the switch to C++ 11 and more.

[jcarpent]

Why do say that hpp-fcl is C++11? This is not the case on my side.
I don't think that we need to go towards std::shared_ptr yet. For hpp-fcl 2.0, we already discussed a bit with @jmirabel on some improvements which may simplify hpp-fcl while also improving the whole performances.
See https://github.com/humanoid-path-planner/hpp-fcl/projects/1

[jcarpent]

hpp-pinocchio is not compatible with pinocchio v2.6.0, so I can't publish pinocchio v2.6.0

What do you mean exactly?
I do not see any reason why hpp-pinocchio should be not compatible with Pinocchio 2.6.0 as there is no API or ABI breaching coming from Pinocchio.

[nim65s]

hpp-fcl is not yet C++11 only.

In this issue, I am suggesting to change this, because this would be easier to switch all HPP at once, instead of having boost::shared_ptr in "hpp-fcl" and std::shared_ptr in "all HPP but hpp-fcl".

This would require less work to get everything working again, and will be faster.

And anyway, I guess we'll want to switch to std::shared_ptr in hpp-fcl at some point. C++11 is 10 years old now. So from my point of view, the question is "What are we waiting for", given that waiting more would require more work.

For hpp-pinocchio, cf. humanoid-path-planner/hpp-pinocchio#146

[jcarpent]

For me, this will break compatibility with Pinocchio, because hpp-fcl objects are exposed in the current API of Pinocchio.
The only solution on my side will be to wait for Pinocchio 3.x which will make C++11 the default C++ version with template generation support.

My veto is to wait for Pinocchio 3.x. We can discuss in private to see what is missing to launch Pinocchio 3.x

[nim65s]

ok, let's synchronize this with Pinocchio 3 then.

[jcarpent]

Thanks @nim65s for your compliance.

[jmirabel]

#256 raised the C++ 11 question again. It has been 9 months since this discussion was started. A few questions:

• What prevents future 2.x release of pinocchio to be C++ 11 ?
• Anyone else has an opinion on this ?

[jcarpent]

• What prevents future 2.x release of pinocchio to be C++ 11 ?

As already discussed, starting from Pinocchio 3.x.

[jmirabel]

Is there any technical point that for keeping c++98 for new 2.x versions ? This reason has frozen this discussion for 9 months and there is no clear timeline for pinocchio 3 release. I am not putting pressure on pinocchio 3. I just think that, if there is no technical reason to keep the 98 compatibility, then at some point we may decide to move forward anyway.

[jcarpent]

For, me the main reason is downstream users of Pinocchio who have to stick to C++98.
I don't want to abandon them at this stage and prefer to prepare the spirit with Pinocchio 3.x
This could be the same for hpp-fcl, with version 2.x and more, compliant with C++14 and more.