Conversation
|
There are bugs in Travis because I forgot to update one part of the code that rely on the PLY reader. I'll fix it. |
|
Bugs should be fixed. |
|
Note (cc @lrineau): we should install LASlib on some of the platforms of the testsuite so that this new feature is tested properly (we now have a dedicated fork that should be easy to install). |
|
@sgiraudot @maxGimeno You should work together and propose a pull-request to some docker images at https://github.com/CGAL/cgal-testsuite-dockerfiles. |
|
And you (Simon&Maxime) should also look with @janetournois for Windows. |
| # It assumes that find_package(LASLIB) was already called. | ||
|
|
||
| add_definitions(-DCGAL_LINKED_WITH_LASLIB) | ||
| add_definitions(-DCGAL_LINKED_WITH_LASLIB -DUSE_AS_DLL) |
There was a problem hiding this comment.
Where does that name USE_AS_DLL come from? It should have a kind of namespacing: LASLIB_USED_AS_DLL, or
CGAL_USE_LASLIB_AS_DLL...
There was a problem hiding this comment.
It's from LASlib.
We forked it, so I could change it, but I'd rather change as little things as possible in the library. Maybe I can add #define USE_AS_DLL inside the CGAL code just before including the LASlib header and #undef right after (there is exactly one inclusion of LASlib). What do you think?
There was a problem hiding this comment.
If the CMake command was a target_compile_definitions, just for a specific target, that would be fine. But here the line add_definitions(-DCGAL_LINKED_WITH_LASLIB -DUSE_AS_DLL) will add -DUSE_AS_DLL to all compilation commands of the CMake project.
If that is not too complicated to hide that macro in just one header of CGAL, I would prefer.
|
Sosno should be ready to test this PR on visual studio 2015. |
|
| ENV LASLIB_LIB_DIR | ||
| ) | ||
|
|
||
| if(LASLIB_LIBRARIES) |
There was a problem hiding this comment.
If LASLIB_LIBRARIES evaluates to TRUE, but not LASZIP_INCLUDE_DIR, there is a CMake error.
On my machine, there is a file /usr/lib64/liblas.so, from the package liblas-devel-1.8.0-13.fc25.x86_64, that corresponds to http://www.liblas.org, and LASLIB_LIBRARIES is automatically set to /usr/lib64/liblas.so, but LASZIP_INCLUDE_DIR remains NOTFOUND.
There was a problem hiding this comment.
The main problem here is that libLAS != LASlib, although both libraries use the same filename… so anyway, there's no way the code will compile, even if we add LASzip. I'm not sure at all how we can handle that.
There was a problem hiding this comment.
I think LASLIB_FOUND should not be set to true if LASLIB_INCLUDE_DIR is not-found. That would probably solve the issue I have had.
|
I added the test for |
|
That works correctly now on my machine: instead of a CMake critical error, I just have the message
as expected. |
|
It seems that PR change the API that was introduced in CGAL-4.8 by Small_Features/Point_set_processing_ply_io and modified in CGAL-4.9 by Small_Features/Extend_PLY_point_set_reader. What is the upgrade path for users of that documented API? If that is a simple breakable, then the |
|
Good point. I've updated |
Installation/changes.html
Outdated
| points with properties is modified for unification with LAS | ||
| (see <code>CGAL::read_ply_points_with_properties()</code>). A new | ||
| function to write PLY with properties is provided | ||
| (<code>CGAL::write_las_points_with_properties()</code>).</li> |
There was a problem hiding this comment.
I think you have swapped the two names of functions CGAL::read_ply_points_with_properties()/CGAL::write_las_points_with_properties() in the description.
|
The Travis error will remain: https://travis-ci.org/CGAL/cgal/jobs/238376800#L4070 The two files should include: #include <CGAL/license/Point_set_processing_3.h> |
|
Alright, I fixed both problems (error in |
|
Doxygen error: https://cgal.geometryfactory.com/CGAL/Manual_doxygen_test/CGAL-4.11-Ic-35/Documentation.log |
… ICC" We have to check `__GXX_EXPERIMENTAL_CXX0X__`. The issue in Boost-1.53 is that the other macro `__GXX_EXPERIMENTAL_CPP0X__` is checked instead.
1807376 to
56919f9
Compare
|
The static assertion should be fixed and the branch is rebased. |
|
Everything looks good except for the remaining warning on Windows DLL. I deactivated it in the code: it comes from LASlib and I'd rather change our fork of LASlib as little as possible (and this warning seems to require many additions of the export DLL mechanism). Anyway, we already deactivated a warning from GCC for the inclusion of LASlib… |
|
The remaining error on Windows should be fixed. |
Summary of Changes
Input/output functions for LAS format (+ updated functions for PLY).
Release Management
Point_set_processing_3