5.0 KiB
Minutes from 27th OpenVDB TSC meeting, September 12th, 2019, (GMT)
Attendees: Nick A., Dan B., Jeff L., Ken M., Peter C.
Additional Attendees: Bruce Chernia (Intel), Dan Elliott (Weta)
Agenda:
-
Quorum
-
Secretary
-
Latest on our release plans, specifically v6.2
-
Changes to unit-tests (Ken) 4b) Staggered Interpolation
-
SPDX Update
-
CLA Update
-
OP_Operator::getVersion format (PR502)
-
OpenVKL
-
Renderman and H18
-
Anisotropic Surfacing
-
Next Meeting
-
Quorum was confirmed.
-
Secretary was Dan Bailey
-
Latest on our release plans, specifically v6.2
Jeff would like a 6.2.0 release within the next ten days so it can be integrated into Houdini 18.0 before the upcoming release. Peter has agreed to do this release, but would like us to move to a model where others on the TSC can also do releases. Primary barrier is improving the way Doxygen gets released as currently this is a manual, offline process and would be better if automated. Dan mentioned that there are a number of other admin tasks that are still outstanding of which this is one.
Outstanding items to complete before we release:
- PR502 - Dan to finish, see discussion in (7).
- PR504 - Nick to do more testing, affects Windows builds, may not make the cut for 6.2.
- PR506 - Nick to merge.
- CMake Deprecation Warnings - Nick to add a few deprecation warnings to CMake.
- Changes to unit-tests (Ken)
Ken would like to see us move to Google Test over CppUnit. More modern C++ unit testing framework actively developed by Google. Various useful features such as parallel testing, running a test until failure, less finicky over type mismatches, etc. Peter concerned about adding a new dependency, Nick suggested Boost Test as an alternative. Some discussion about our policy regarding Boost, general feeling that we have been trying to migrate towards using standard library as functionality becomes available but no plan to try and accelerate this. It is deemed more important to reduce the use of Boost in the core library than the unit tests. It was also suggested that we should look at other open-source projects and which unit test framework they use.
All agreed that we would benefit from the increased feature set provided in Google Test, but less certainty about whether the migration would be worth the effort. Ken to share an example as a test case of how complicated the migration might be.
Dan to share a link to the SonarCloud code coverage so that all can get an understanding of areas of the codebase that might be lacking in coverage.
4b) Staggered Interpolation
Ken mentioned that in writing a simple fluid solver, he ran into issues with efficiency of interpolation of staggered grids, Nick concurred. It is currently redundantly interpolating in axes that are not used and there is no stencil caching as recently discussed in regards to sharpening. Ken to look at contributing an improvement here.
- SPDX Update
John Mertic not on the call, Dan to reach out to find out if there has been any progress here.
Update: John Mertic and Steve Winslow had a conference call with Andrew Pearce. Dreamworks have some concerns on the impact of removing the language from the license moving to the SPDX license. John and Steve are exploring options to resolve the non-standard license language while at the same time address Dreamworks' concerns.
- CLA Update
Ken, Jeff and Dan have not completed the digital CLAs. This is holding up our ability to accept external contributions. Dan is still awaiting Lucasfilm/Disney legal to proceed. Ken and Jeff to follow up on their end. Ken to look at signing the individual CLA.
- OP_Operator::getVersion format (PR502)
Peter has suggested both Houdini and VDB versions to be included in the version string for possible future use and no objections to this. Question over whether there is a benefit to align with existing version mechanism as OP_Operator::getVersion is a public function. Jeff said that lots of hip file nodes such as HDAs use custom strings which cannot be parsed so no real value in attempting to do so. Jeff proposed "vdb6.1.0 houdini18.0.222" as the format. All in agreement. Dan to make that change in time for inclusion in a 6.2.0 release.
- OpenVKL
Bruce to discuss with Greg about the possibility of the TSC receiving an OpenVKL update from Intel around October. Intel's plan is to incorporate VDB in Q4, OpenVKL not intended to replace OpenVDB.
- Renderman and H18
Dan mentioned a discussion being had offline with the Renderman group. Hope that the developers will be able to join us in a future call to discuss and may even be able to contribute to the project in the future.
- Anisotropic Surfacing
Nick looking at open-sourcing this tool. Jeff not certain of the exact implementation but confirmed that the Particle Fluid Surface SOP is the primary tool for surfacing particles provided by Houdini.
- Next Meeting
Next planned meeting is:
September 26th 2019. 3pm-4pm EDT (GMT-4). Option to hold one on September 19th, depending on the status of the 6.2.0 release.