On Wed, Feb 26, 2025 at 10:55:49PM +0000, Johannes Thyssen Tishman wrote: > 2025-02-26T20:21:45+0100 Theo Buehler <t...@theobuehler.org>: > > On Wed, Feb 26, 2025 at 05:54:44PM +0000, Johannes Thyssen Tishman wrote: > > > Please find below an update for graphics/pcl to version 1.15.0. > > > > > > Most patches were dropped as they were merged upstream. I checked all > > > SHARED_LIBS with check_sym for bumps, but I'd appreciate if someone > > > could help me double check. All tests are still passing. > > > > The bumping looks correct purely from looking at the check_sym output. > > There are too many headers to be confident about function signatures > > and struct changes, though: there was a lot of churn. Moreover, all > > these libraries depend on libpcl_common, which got a major bump. > > > > In practice, a binary linked against, say, libpcl_kdtree from pcl 1.14.1 > > will no longer work with pcl 1.15.0 despite libpcl_kdtree not having any > > dynamic export changes itself. So there is no real benefit from avoiding > > a bump for libpcl_kdtree. > > > > In short, I would recommend doing the same you did with vtk and just > > give all the libraries the same version, in this case 1.0. This will > > save you a lot of time and headache and I see no downside to doing > > that, especially since there is no direct consumer of this port (yet). > > Thanks for double checking tb. I bumped them all now as done with vtk. > New diff below.
ok tb > > On aarch64 tests pass except for two that time out: > > > > The following tests FAILED: > > 83 - io_io (Timeout) > > 99 - a_recognition_ism_test (Timeout) > > Do you have a more verbose log of these? I don't have an aarch64 machine > to test. Not really interesting. I have a local diff in my cmake.port.mk that reduces the annoyingly long test timeout from 120s to 20s and these two tests take longer than that on my m1. All tests pass if I reinstate the default. > > One question: what happened to the dep on graphics/freeglut? > > > > If I see correctly, there's grabcut_2d that would depend on it > > but it is only built if GLUT_FOUND is set and nothing seems to be > > setting that. > > Nice catch! I intended to prepend devel/cjson to the list (alphabetical > order) and removed graphics/freeglut by mistake. freeglut is searched > for due to WITH_OPENGL being set. Yes. grabcut_2d still isn't built.
