search

jrl-umi3218 / sch-creator

Creation of convex hulls volumes
BSD 2-Clause "Simplified" License
1 stars 3 forks source link

License #4

Open mehdi-benallegue opened 3 years ago

mehdi-benallegue commented 3 years ago

Is there a reason for which this package is GPL-v2 ?

For sch-core, it is related to penetration depth computations, but as far as I know we have only our code here (Adrien, Amélie, and me mostly).

Could we change it something more modern?

kheddar commented 3 years ago

I believe that @gergondet already replaced the depth computations by our code as per a request of a Harvard company of surgery that used the code. If this is confirmed that Pierre can change this.

mehdi-benallegue commented 3 years ago

I believe that @gergondet already replaced the depth computations by our code

That is good news! But it concerns sch-core and not this specific package (sch-creator). This one contains only our code as far as I can tell. I asked Adrien and he agrees. So if nobody sees a reason to keep this license I can update it.

kheddar commented 3 years ago

Fine but please know that the depth computation (because you mention it) is replaced by own code and the whole distance/collision computation package can have the appropriate license.

mehdi-benallegue commented 3 years ago

I just checked again and AFAIK what he did is to design a BSD build of the library that does not include penetration depth, so we still don't have our code for that.

Anyway, this involves another library, I am cleaning this one now. Thanks a lot for the reply !

gergondet commented 3 years ago

To answer @mehdi-benallegue original question

Is there a reason for which this package is GPL-v2 ?

AFAIK this repository is not used anymore and we only use the qhull-generated convex hulls (via https://gite.lirmm.fr/multi-contact/robot_script/-/tree/blender2.8)

To confirm/clarify the situation with sch-core, there is two versions with a different license and inter-penetration depth computation:

See the original PR related to this: https://github.com/jrl-umi3218/sch-core/pull/22

mehdi-benallegue commented 3 years ago

AFAIK this repository is not used anymore

Is that because of STP-BV building issues ?

We are currently working with a student to make the algorithm faster and more robust. So I want to revive this repo.

gergondet commented 3 years ago

Is that because of STP-BV building issues ?

I'm not sure what issues you are referring to here?

From what I recall, Joris stopped using STP-BV through his PhD as:

  1. mostly https://github.com/jrl-umi3218/sch-core/issues/4
  2. convex hulls were ok (i.e. fast enough)

We are currently working with a student to make the algorithm faster and more robust. So I want to revive this repo.

Ok. FYI, the blender addon - and probably the blender-related scripts - in this repo are outdated since Blender API was completely changed in version 2.8, the repository I have linked above has an up-to-date version

mehdi-benallegue commented 3 years ago

I'm not sure what issues you are referring to here?

Being unstable and slow (the building failed too often and finding the initial triangle is O(n^4) complexity).

Thank you for the reply. The issues you are referring to are mostly fixed (except that STP-BV is still a bit slower than convex-hulls). So I think it is worth giving the last shot on them (and at least a paper).

Thank you for the advice for the scripts, I will adopt the ones you pointed out.