4.2 KiB
Minutes from 106th OpenVDB TSC meeting, September 28th, 2021, (EDT)
Attendees: Ken M., Dan B., Nick A., Jeff L., Andre P.
Additional Attendees: JT Nelson (Blender), Richard Jones (DNeg), Johannes Meng (Intel)
-
Confirm Quorum
-
Secretary
-
CI Issue that OpenVDB is having
-
Forum
-
NanoVDB
-
SIGGRAPH Feedback
-
OpenVDB v9
-
OpenEXR v3
-
Mathematica
-
Next Meeting
-
Confirm Quorum
Quorum is present.
- Secretary
Secretary is Andre Pradhana.
Nick mentions that there is a lot of discussion on NanoVDB. Ken will take a look.
- CI Issue
Dan mentions that there are some CI issues with the Houdini CI. This problem is due to not setting the RPATH in pyOpenVDB component; it's because when you find package BLOSC, it is finding it from the Houdini DSO Lib directory. But that find package call is only happening inside the OpenVDB core library cmake, which is not happening at the right level. Nick will take a look at this, especially if the find package calls makes sense to live at the top level.
- Forum
Nick mentions that there are a few NanoVDB related issues/requests on the github issues.
- NanoVDB
Ken created a PR. He did what we talked about before. It's trying to merge into feature/nanovdb, instead of master. Nick says that it looks good, but he mentions that more care is needed from the fact that the NanoVDB viewer is included in that feature branch. Ken mentions that he and Jeff see that the viewer offers an advantage in that it shows people how they can adopt NanoVDB in their project and that it serves as an example project. The viewer is working out-of-the box in its current version, but since the build-system for NanoVDB is going to change (by the fact that we are trying to make the build system for NanoVDB and the rest of the OpenVDB projects to be more aligned), he sees that making the build-system to also work with the viewer is a tall order.
At the end, we decided to make a new PR based on feature/nanovdb_merge_candidate and to leave feature/nanovdb alone. We then go back to fix feature/nanovdb.
- SIGGRAPH
Ken says that it's a very good feedback from SIGGRAPH regarding the OpenVDB course. There is one comment he finds really interesting: someone says that the presentation is great, but he wants to know how to use the library. The commenter is suggesting a course on programming. Nick mentions that Dan's presentation is related to this request. Ken mentions that Dan's course is more geared towards an expert's perspective on how to accomplish something efficiently, instead of answering, "How do I do anything?" Jeff mentions that it is similar to a course on how to use an API. Nick, Dan, and Jeff suggest that providing an example in the library.
OpenVDB has a lot of unit tests that acts as examples. NanoVDB comes with a lot of examples. Ken thinks that we should do the same for OpenVDB. Ideally, we should update the cook-book.
- OpenVDB v9
The root-offset. Ken has not made any progress on that. He will look at it.
Dan talks about explicit template instantiation, this is PR-1182. It is working on Windows. The cmake is all there. He did one test with Activate.h. He is hoping to get sign-off on this.
Dan gives an overview of the PR by sharing screen. He needs to remove inline statements from the tools. It boils down making a single macro call. Jeff suggests that we need to have undef in the define macro in Activate.h.
Nick makes a comment that cmake should generate the .cc files, instead of make. We can use cmake files method to write the files. Nick suggests Dan to take a look at execute_process command in cmake. The plan is to put it in v9. Once this PR is approved, Dan will make another PR to invoke this mechanism for all the tools.
- OpenEXR 3
There is a lot of back-on-forth with the OpenEXR team. Nick mentions that it looks fine. The more annoying to have a hash-define.
- Mathematica
Ken has a meeting with a company that is doing 3D-printing. They have integrated OpenVDB into Mathematica. They are still stuck on the CLI part, but once they sign that, they want to contribute back. It may end up as a module, like Maya or Houdini. Ken wants to invite them.
- Next Meeting
Next meeting is October 5th, 2021. 12pm-1pm EST (GMT-5).