Commit Graph

228 Commits

Author SHA1 Message Date
Joydeep Biswas 88e08cfe71 Mixed-precision Iterative Refinement Cholesky With CUDA
* Created a new class CUDADenseCholeskyMixedPrecision, which performs
  Cholesky factorization and solving in single (fp32) precision, and
  optionally performs iterative refinement.
* Added CUDA kernels for mixed-precision solve operations
* Added more detailed timing information to the FullReport about Schur
  elimination, reduced system solves, and back-substitution.

Some test performance numbers follow.
All tests were performed on an Ubuntu 20.04 desktop with an
Intel Core i9-9940X CPU and Nvidia Quadro RTX 6000 GPU.

Tests were launched as:
./bin/bundle_adjuster --input (problem_file) \
    --num_iterations 20
    --num_threads 28
    --linear_solver dense_schur
    --dense_linear_algebra_library (cuda|lapack)
    [--mixed_precision_solves]

==================================================
problem-21-11315-pre.txt
==================================================

--------------------------------------------------
Cuda Mixed Precision
--------------------------------------------------
Cost:
Initial                          4.413239e+06
Final                            3.037864e+04
Change                           4.382861e+06
  Linear solver                      0.250703 (14)
  ├ Schur eliminate                  0.234025 (14)
  ├ Reduced solve                    0.006643 (14)
  └ Backsubstitute                   0.006598 (12)

--------------------------------------------------
Cuda
--------------------------------------------------
Cost:
Initial                          4.413239e+06
Final                            3.037864e+04
Change                           4.382861e+06
  Linear solver                      0.257517 (12)
  ├ Schur eliminate                  0.233518 (12)
  ├ Reduced solve                    0.010621 (12)
  └ Backsubstitute                   0.007124 (12)

--------------------------------------------------
Lapack (OpenBLAS)
--------------------------------------------------
Cost:
Initial                          4.413239e+06
Final                            3.037864e+04
Change                           4.382861e+06
  Linear solver                      0.332349 (12)
  ├ Schur eliminate                  0.274748 (12)
  ├ Reduced solve                    0.015966 (12)
  └ Backsubstitute                   0.034192 (12)

==================================================
problem-257-65132-pre.txt
==================================================

--------------------------------------------------
Cuda Mixed Precision
--------------------------------------------------
Cost:
Initial                          2.456242e+07
Final                            9.677593e+04
Change                           2.446565e+07
  Linear solver                      1.332367 (20)
  ├ Schur eliminate                  1.021365 (20)
  ├ Reduced solve                    0.195472 (20)
  └ Backsubstitute                   0.075582 (20)

--------------------------------------------------
Cuda
--------------------------------------------------
Cost:
Initial                          2.456242e+07
Final                            9.677547e+04
Change                           2.446565e+07
  Linear solver                      1.810176 (20)
  ├ Schur eliminate                  1.012862 (20)
  ├ Reduced solve                    0.678704 (20)
  └ Backsubstitute                   0.083925 (20)

--------------------------------------------------
Lapack (OpenBLAS)
--------------------------------------------------
Cost:
Initial                          2.456242e+07
Final                            9.677547e+04
Change                           2.446565e+07
  Linear solver                      2.376273 (20)
  ├ Schur eliminate                  0.987613 (20)
  ├ Reduced solve                    1.043873 (20)
  └ Backsubstitute                   0.310402 (20)

==================================================
problem-744-543562-pre.txt
==================================================

--------------------------------------------------
Cuda Mixed Precision
--------------------------------------------------
Cost:
Initial                          1.434881e+08
Final                            1.546895e+06
Change                           1.419412e+08
  Linear solver                     27.010088 (20)
  ├ Schur eliminate                 24.362433 (20)
  ├ Reduced solve                    1.428542 (20)
  └ Backsubstitute                   0.814266 (20)

--------------------------------------------------
Cuda
--------------------------------------------------
Cost:
Initial                          1.434881e+08
Final                            1.546895e+06
Change                           1.419412e+08
  Linear solver                     32.342513 (20)
  ├ Schur eliminate                 24.638819 (20)
  ├ Reduced solve                    6.492090 (20)
  └ Backsubstitute                   0.802184 (20)

--------------------------------------------------
Lapack (OpenBLAS)
--------------------------------------------------
Cost:
Initial                          1.434881e+08
Final                            1.546895e+06
Change                           1.419412e+08
  Linear solver                     34.152224 (20)
  ├ Schur eliminate                 24.183723 (20)
  ├ Reduced solve                    8.784413 (20)
  └ Backsubstitute                   0.795044 (20)

Change-Id: I178887e776d8f4a1e8abb99bbc205bf8c278bf79
2022-07-13 06:55:31 -05:00
Alex Stewart 5de77f399e Fix reporting of METIS version
- Also fixes behaviour of EIGENMETIS option to match that of the other
  CMake dependency options, and ensure that its value aligns exactly
  with whether Eigen support for METIS will be compiled into Ceres.

Change-Id: Ifbf6f5d82b9ba89a156673eb6042519a985e6b04
2022-06-22 19:20:40 +01:00
Sergiu Deitsch 0c88301e66 Provide optional METIS support
* Split `CERES_NO_METIS` into two defines: `CERES_NO_PARTITION` and
  `CERES_NO_METIS`. The former refers to METIS support in SuiteSparse,
  the latter to the Eigen's MetisSupport module. This enables the use of
  sparse matrix reordering independent from SuiteSparse.
* Run Linux, macOS, and macOS Github workflows with METIS enabled
  SuiteSparse.

Fixes #808

Change-Id: I5076b7e1268d32cc3e7e56650edcbaf7fb3b59ce
2022-06-22 16:46:02 +00:00
Sameer Agarwal 2335b5b4b7 Remove support for CXSparse
Eigen provides all the functionality that we need from CXSparse
with a more liberal license.

I will update the documentation in a follow up CL.

Change-Id: I0b9fd8be3c27754cc2986cc0e06595c8b3fdec0b
2022-05-27 09:20:21 -07:00
Sameer Agarwal 12263e2830 Make the min. required version of SuiteSparse to be 4.5.6
With this change we can drop the complicated/conditional handling
around CAMD and assume that it is always available.

Change-Id: I93e1da676fb75817f79824b8b2b6549d03f278b0
2022-05-16 12:48:43 -07:00
Evan Levine f1414cb5bd Correct spelling in comments and docs.
Change-Id: Iad9a0599d644d3b3cd54244edaf64d408cb1308e
2022-04-24 21:40:13 -07:00
Sergiu Deitsch 817f5a0688 Switch to imported SuiteSparse, CXSparse, and METIS targets
These changes allow the use of a SuiteSparse CMake package from
https://github.com/sergiud/SuiteSparse that allows native compilation of
SuiteSparse using CMake on a variety of platforms Packages generated
using official SuiteSparse makefiles can still be used without
modifications. The find module remains agnostic to specific CMake
package implementation.

CMake packages have the advantage that they are self-contained and
relocatable. The latter is particularly useful in cross-compilation
scenarios.

Fixes #728

Change-Id: I089d5c6f87c05b1530a5ab9a36dff2fcbe82d13d
2022-03-03 21:26:45 +01:00
Sergiu Deitsch f0851667be Fix MSVC compilation errors
Disable the definition of `min`/`max` macros by defining `NOMINMAX`
and prevent macro substitution in the public interface.

Also, quiet floating-point comparisons are defined as template functions
by the MSVC STL which causes compilation errors due to ambiguities in
resolving the template parameter types.

Fixes #668
Fixes #716
Fixes #718

Change-Id: I5fe7832a6a3a7ad0421a2557527528c34b88e9c7
2022-02-27 13:24:42 +01:00
Sergiu Deitsch 46b3495a4f Standardize path handling using GNUInstallDirs
Fixes #649

Change-Id: Iefb3e7be69a04474db0fc768b47f4f3636c8a587
2022-02-25 23:59:34 +00:00
Sergiu Deitsch 09ec4997fa Cleanup examples
Remove logic invoked based on obsolete variable definitions. Use new
(explicit) target_link_libraries syntax to link binaries against
dependencies. Do not rely on prior knowledge about the compiler for
specifying flags and system libraries but instead directly test their
presence to be more robust.

Change-Id: I76e0d10fae6eba4b343048e4404f0a9b08c7cb6c
2022-02-18 00:53:44 +01:00
Sergiu Deitsch 57ec9dc92e Do not enforce a specific C++ standard
Current language standard selection mechanism is quite convoluted and
the logic for propagating a specific C++ language standard (which cannot
be fully enforced anyway) is unnecessarily complicated. Instead
communicate the minimum required C++ standard and let CMake handle the
requirement. This allows the compiler and consumers to use newer
language revisions if available. The language standard used to compile
Ceres solver can still be set via `CMAKE_CXX_STANDARD`.

Move to CMake 3.10 which is supported by Ubuntu 18.04 LTS and simplify
the corresponding language features request.

Change-Id: Ib1f95e2bc5f06bf1275a62565d303eb0f114d127
2022-02-15 21:14:29 +01:00
Sergiu Deitsch 99698f0535 Fix Apple Clang weak symbols warnings
Change-Id: I71eba56ca37060c83d05ff14d6f9bfaef61e9493
2022-02-15 21:00:22 +01:00
Sameer Agarwal aff51c907d Revert "Do not enforce a specific C++ standard"
This reverts commit d839b77928.

Reason for revert: Breaks the BUILD on macOS

Change-Id: I49f448ed4942cc88df9e4ad3b78d56f6be04351d
2022-02-15 19:17:02 +00:00
Sergiu Deitsch d839b77928 Do not enforce a specific C++ standard
Current language standard selection mechanism is quite convoluted and
the logic for propagating a specific C++ language standard (which cannot
be fully enforced anyway) is unnecessarily complicated. Instead
communicate the minimum required C++ standard and let CMake handle the
requirement. This allows the compiler and consumers to use newer
language revisions if available. The language standard used to compile
Ceres solver can still be set via `CMAKE_CXX_STANDARD`.

Move to CMake 3.10 which is supported by Ubuntu 18.04 LTS and simplify
the corresponding language features request.

Change-Id: Id3526c69990315289e7ea0fbcdeaa6af79d24d03
2022-02-15 19:10:51 +01:00
Joydeep Biswas 7d2e4152ec Add support for dense CUDA solvers #2
1. Add CUDADenseQR & tests.
   CUDADenseQR uses the cuSolverDN LAPACK implementation
   of QR factorization. A key limitation, however, is that
   this solver does not perform singularity checking --
   this is because cuSolverDN does not have a trtrs
   implementation; we instead use cuBLAS' trsv for
   backsubstitution.
2. All CPU -> GPU memory transfers are now async, and both
   CUDADenseQR and CUDADenseCholesky explicitly manage their
   own streams for async operations.
3. Simplified CUDADenseCholesky to only use the legacy 32-bit
   cuSolverDN API.

Change-Id: I2a9b7b65469658ddfe33b5b2a3892c8744d6e437
2022-02-14 15:22:39 -06:00
Sergiu Deitsch f90833f5fa Simplify symbol export
Currently, the logic for exporting symbols is rather complicated: when
tests are enabled internal symbols are exported in addition to the
public symbols. Such logic causes several problems. (1) Test binaries
link against a Ceres build that is different from the final release
since fewer optimizations are applied if more symbols are exported. (2)
Also, some toolchains hide symbols by default breaking the existing
logic eventually causing linker errors.

Since internal symbols are not intended to be used outside of the
project, we can compile them into object files and use exactly the same
binary code both for the final build and the tests without relying on
conditionals.

By default, all symbols are now hidden unless annotated as public.
Internal symbols are explicitly marked as not being exported in case
users chose not to hide symbols by default.

Change-Id: I589dd10be2f6f438508783cf99d141af0120057b
2022-02-14 20:19:08 +01:00
John Harrison e0fef6ef05 Add cmake option ENABLE_BITCODE for iOS builds
Change-Id: Ib56c711c626044754c1c3a397d7c4648c4ec089d
2022-02-13 15:54:24 +00:00
Sergiu Deitsch 708a2a7233 Silence LocalParameterization deprecation warnings
Suppress warnings in all Ceres targets without leaking suppressions into
user code (fixes #750).

Change-Id: I3aa1063cb0aae547865a15311ac286ef6d81a955
2022-02-10 20:55:25 +00:00
Sameer Agarwal 0141ca090c Deprecate LocalParameterizations
Add [[deprecate]] notices to everything LocalParameterization
related.

Make sure that Ceres can be compiled without triggering
deprecation warnings.

Update the documentation:

a. Add deprecation notices.
b. Document interaction between LocalParameterization and Manifold
   coexisting in the Problem.
c. Add documentation for Manifold(s)

Change-Id: Ie4ad48963c83fded86e533c8c60561af402fbaff
2022-02-10 06:36:47 -08:00
Joydeep Biswas 36d6d86908 Add support for dense CUDA solvers #1
1. Add CUDADenseCholesky64Bit, CUDADenseCholesky32Bit, & tests.
   CUDADenseCholesky32Bit uses the legacy versions of potrf/potrs
   in cuSolverDN, while CUDADenseCholesky64Bit uses the new 64-bit
   versions available since Cuda 11.1. The legacy versions are
   provided since some platforms such as the Nvidia Jetsons only
   support Cuda 10.2.
2. Expose CUDA as a new option under DenseLinearAlgebraLibraryType.
   The relevant option to string and string to option helper functions
   are modified accordingly.
3. Add cuda as a dense_linear_algebra_library option in bundle_adjuster
   to demonstrate the use of the new CUDA option.

Change-Id: I23615e1d301df5185ed646b3e33ee802508dae86
2022-02-07 19:26:29 -06:00
Sameer Agarwal 125a0e9be5 LocalParameterization -> Manifold #1
Manifolds are now part of the public API and co-exist
with LocalParameterizations.

1. Add Manifolds to the Problem API.
   a. AddParameterBlock(double*, int, Manifold*)
   b. SetParameterization(double*, Manifold*)
   b. GetManifold(const double*)
   c. HasManifold(const double*)

2. Internally Ceres now only uses Manifolds. When the user uses
   a LocalParameterization, it is wrapped in a ManifoldAdapter.

3. To preserve the API semantics while keeping the internals clean
   we need a new map in ProblemImpl which stores the association
   between parameter blocks and local parameterizations. This
   is temporary, it will go away once this transition is complete.

4. There are NO algorithmic changes, as in we are not using
   any of the expanded interface of the Manifold objects yet.
   That will come later.

5. All tests that use LocalParameterization have been duplicated
   to use Manifolds, and when this transition is complete the
   LocalParameterization based tests will be deleted.

6. Public documentation for the API has been updated. Deprecation
   notices to the documentation as well as C++ annotations will come
   later.

7. Similar changes have been made to GradientProblem.

Change-Id: I8e03c8ced6e141876ef3eca5740c113afa788f0c
2022-01-18 15:59:45 -08:00
Alex Stewart 8d3e64dd5e Use modern-style Eigen3 CMake variables
Change-Id: I4c14aa54552b3f718dff90eb397cb030aeb6ecd4
2021-11-16 19:51:06 +00:00
Sumit Dey 7de561e8e8 Fix dependency check for building documentation
Build with documentation fails if the required 'sphinx rtd theme' is
not available. Check if dependency is installed before building with
documentation.

Add Python3 as requirement for building documentation.

Change-Id: I5edc5d7374864990e625a5efb358f5a23b3c50fe
2021-07-31 12:21:16 +02:00
Florian Berchtold 323c350a66 fix Eigen3_VERSION
Change-Id: Ida3796ec9d5732b284973dbbf0a395f9c5b19a7b
2021-03-03 19:30:08 +01:00
Taylor Braun-Jones 3f6d273676 Unify symbol visibility configuration for all compilers
This makes it possible to build unit tests with shared libraries on MSVC.

Change-Id: I1db66a80b2c78c4f3d354e35235244d17bac9809
2020-10-15 16:56:07 -04:00
Alex Stewart 4e69a475cd Fix potential for mismatched release/debug TBB libraries
- Protect against the case when the user has multiple installs of TBB
  in their search paths and the first install does not contain debug
  libraries.  In this case it is possible to get mismatched versions
  of TBB inserted into TBB_LIBRARIES.
- Also suppresses warning about use of TBB_ROOT on modern versions of
  CMake due to CMP0074.

Change-Id: I2eaafdde4a028cbf6c500c63771973d85bc4723d
2020-09-03 19:55:58 +01:00
Alex Stewart 29fb08aeae Use CMAKE_PREFIX_PATH to pass Homebrew install location
- Passing HINTS disables the MODULE mode of find_package() which
  precludes users from creating their own find modules to provide
  Ceres' dependencies.

Change-Id: I6f2edf429331d13fe67bf61ac4b79d17579d9a57
2020-09-01 10:23:31 +01:00
Alex Stewart 7d3ffcb423 Remove forced CONFIG from find_package(Eigen3)
- Ceres will fail to configure if Eigen3::Eigen target is not found, and
  the minimum required Eigen version specified (3.3) exports Eigen as
  a CMake package and this is reflected in the default Ubuntu 18.04
  packages.
- This permits users to specify their own Eigen3 detection should they
  choose to do so, but they must do so via an imported target.

Change-Id: I5edff117c8001770004f49012ac1ae63b66ec9c1
2020-08-04 21:16:25 +00:00
Alex Stewart aa1abbc578 Replace use of GFLAGS_LIBRARIES with export gflags target
- As our minimum required version of gflags (2.2) exports itself as
  a CMake package and this is the case for the default 18.04 package
  we can use the gflags target directly.
- Replaces forced use of CONFIG in find_package(gflags) with a check
  that the gflags imported target exists to avoid ambiguity with
  libgflags if installed in a default location.  This permits users to
  override the gflags detection should they so choose, provided that
  they do so via an imported target.
- Also removes some previously removed legacy GLAGS_ vars from the
  installation docs.

Change-Id: I015f5a751e5b22f956bbf9df692e63a6825c9f0d
2020-08-04 21:15:51 +00:00
Alex Stewart 7ef83e0759 Update minimum required C++ version for Ceres to C++14
- Removes all workarounds for pre-C++14 versions
- Removes '11' qualifier from C++ threading option and associated
  defines.
- Fix missing inclusion of 'Multithreading' in reported Ceres components
  when C++ threading model is enabled.
- Update Sphinx documentation to specify C++14 as minimum requirement.

Change-Id: I706c8b367b3221e3c4d1a0aaf669a8f9c911e438
2020-05-30 19:15:03 +01:00
Alex Stewart cca93fed63 Bypass Ceres' FindGlog.cmake in CeresConfig.cmake if possible
- If the version of glog detected and used to build Ceres was built with
  CMake (i.e. Config mode) then we now use Config mode directly in
  CeresConfig.cmake and do not install Ceres' FindGlog.cmake module.
- This has the benefit of removing any hard-coded paths from
  CeresConfig.cmake provided that all dependencies were also built with
  CMake.

Change-Id: I85af8a953fd6d300e8bc0cdeb0b3636fec182f68
2020-05-25 17:20:12 +00:00
Alex Stewart a09682f00d Fix MSVC version check to support use of clang-cl front-end
- Raised as issue: #521

Change-Id: Iaea6b43484b90ec8789bda0447c8a90759974ec1
2020-05-25 17:17:42 +00:00
Alex Stewart b70687fcc8 Add namespace qualified Ceres::ceres CMake target
- This reflects modern CMake style, and also provides a measure of
  protection against missing find_package() imports in downstream
  clients resulting in linker errors when 'ceres' matches the compiled
  library and not the imported target.
- The original 'ceres' target remains, as a local imported interface
  target created by CeresConfig for backwards compatibility.

Change-Id: Ie9ed8de9b7059bc0cae1ae5002bb94d8fe617188
2020-05-25 17:16:27 +00:00
Alex Stewart 99efa54bdb Replace type aliases deprecated/removed in C++17/C++20 from FixedArray
- Raised as issue #551
- Also adds C++20 to the set of ALLOWED_CXX_STANDARDS, although this
  will require a version of CMake >= 3.12.

Change-Id: I0f13c72e93a35391fd2d18590b4243a329a2322c
2020-05-24 19:18:38 +01:00
Darius Rueckert e751d6e4f0 Remove AutodiffCodegen
- Remove Codegen files
- Revert Jet and Rotation

Change-Id: I005c5f98f2b6dfa5c7fd88d998b6aa83e47dab60
2020-04-08 10:43:53 +02:00
Darius Rueckert e9eb76f8ef Remove AutodiffCodegen CMake integration
Change-Id: I403597540df8429378336626b8f748b7821fe6f5
2020-04-06 11:11:43 +02:00
Darius Rueckert 90799e29e1 Fix install and unnecessary string copy
- Fix the following issue when running "make install"
  https://github.com/ceres-solver/ceres-solver/issues/527

- Fix error that CeresCodeGeneration.cmake was not found
  after instalation. Issue:
  https://github.com/ceres-solver/ceres-solver/issues/561

- Removes the unnecessary string copy during code generation

Change-Id: I01963d01da6a9c4557aad6f89831647c1a149e38
2020-02-14 13:11:15 +01:00
Darius Rueckert 032d5844c2 AutoDiff Code Generation - CMake Integration
This patch integrates the code generation module into the build
system. All depenendcies are tracked through CMake targets.
Modifying the cost functor will automatically trigger code
re-generation.

All this functionality is defined in the CMake function
ceres_generate_cost_function_implementation_for_functor
in CeresCodeGeneration.cmake. A hello world usage example
is included in examples/CMakeLists.txt.

Change-Id: I23b8b6698d1ea51cf3d788a47afcf39f8c5ce327
2020-02-13 13:57:14 +01:00
Johannes Beck f26f954105 Fix windows MSVC build.
This CL improves the build experience with MSVC:
- It adds the build flag '/bigobj' otherwise the build of
  the unit test fails.
- It adds the flag '/wd4267' to suppress signed / unsigned int
  conversions (size_t to int).
- It removes the use of std::aligned_storage from FixedArray.
  This has been done from the Abseil Team and is in the
  absl::FixedArray. Those changes has been ported to ceres.
  This fixes the alignemnt for older MSVC versions due to a
  bug in the implementation of std::aligned_storage, and
  prevents the use of the macro '_ENABLE_EXTENDED_ALIGNED_STORAGE'
  for newer MSVC versions, which is problematic as it could affect
  user code.
- Fix of the fixed array unit test. Due to the use of std::tuple
  instead of absl::tuple in ceres::internal::FixedArray, the
  unit test needs to reflect that change as well.
- Replaces 'add_definitions' with 'add_compile_options' for
  compiler flags as suggested by the cmake documentation.

Change-Id: I63f08cd6c0a8db8c9931289b909b4deafd75b039
2020-01-07 17:25:54 +00:00
NeroBurner a3696835b4 use CMake function to create CeresConfigVersion
Use the CMake provided module CMakePackageConfigHelpers and its function
write_basic_package_version_file() to write the CeresConfigVersion.cmake
file to be installed.

Use SameMajorVersion compatibility flag.

Change-Id: I353aa8ede1ffee34e58b41371958ed5dedca4104
2019-12-17 18:40:26 +00:00
NeroBurner ba6e5fb4ad Make the custom uninstall target optional
Ceres-solver provides a custom uninstall target to ease the removal of
installed files from the system. This clashes with other projects, that
too provide an uninstall target (like Eigen3).

Related issue: https://github.com/ceres-solver/ceres-solver/issues/451#issuecomment-565571090

Change-Id: Id153830ae20a880d23c7468acb39f55f48a2129a
2019-12-16 19:25:42 +00:00
NeroBurner a548766d14 Use glfags target
Update the usage of Google Flags (gflags) library the same way Glog
updated it [1]. This pushes the minimum required gflags version to
v2.2.0.

Remove the ceres specific define of CERES_GFLAGS_NAMESPACE and directly
use GFLAGS_NAMESPACE defined in gflags/gflags_declare.h [2].

In CeresConfig.cmake the hard coded paths for gflags are ommited.
Instead we rely on the gflagsConfig file to get closer to a relocatable
CeresConfig.cmake.
Furthermore use the find_dependency() [4] cmake function specifically
created for cmake-config files.

This change builds upon the explicit PUBLIC/PRIVATE link change [3].

[1] https://github.com/google/glog/pull/199
[2] https://github.com/gflags/gflags/blob/d9b184bd0026b16bb4c2fded75d56fb2cce50d66/src/gflags_declare.h.in#L43
[3] https://ceres-solver-review.googlesource.com/c/ceres-solver/+/16220
[4] https://cmake.org/cmake/help/latest/module/CMakeFindDependencyMacro.html

Change-Id: I9861a2699f2702bf1a5e99d07863a7e6639b7c39
2019-12-12 14:00:59 +00:00
NeroBurner 33dd469a53 Use Eigen3::Eigen target
Since version 3.3 Eigen provides Eigen3Config.cmake with the imported
target Eigen3::Eigen. [1]
Use this imported target as descibed in Eigen-Wiki [2]

In the CeresConfig file improve relocatability by removing absolute
paths to the compiled dependencies. Instead find the used Eigen3::Eigen.
Furthermore use the find_dependency() [4] CMake function instead of the
find_package() call in CeresConfig.

This commit relies on all targets to be explicitly linked private or
public as done in the the change [3]

[1] https://bitbucket.org/eigen/eigen/pull-requests/257/cmake-imported-target-take-2/diff
[2] https://eigen.tuxfamily.org/dox/TopicCMakeGuide.html
[3] https://ceres-solver-review.googlesource.com/c/ceres-solver/+/16220
[4] https://cmake.org/cmake/help/latest/module/CMakeFindDependencyMacro.html

Change-Id: I44f44a089083f7169bcf430b59775242e4eb72d1
2019-12-12 14:56:00 +01:00
Darius Rueckert c7337154e9 Disable the code generation module by default
After this patch, users have to add -DCODE_GENERATION=ON to the
cmake command, if they want to compile the code generation module.
A warning is printed if they enable code generation.

This informs the users that the code generation is still under
development and should be used with care. Also, we don't break
the master branch immediately if one of the bigger codegen
patches fails to build on some platform.

When this system is finished, we can enable it by default or
remove this option again.

Change-Id: Ib26498f0d5bd8b3c165807ffd774c057c2d21d39
2019-12-05 17:44:07 +01:00
Darius Rueckert c56702aacc Fix installation of codegen headers
Change-Id: I32751ea992ac3658827941c0dda4753da194affb
2019-12-03 10:40:45 +01:00
Alex Stewart a3dde68770 Require Xcode >= 11.2 on macOS 10.15 (Catalina)
- As detailed in a previous CL, Xcode 11.0-1 exhibited a -fstack-check
  bug on 10.15 Catalina that broke alignment for at least AVX
  instructions causing segfaults from within Eigen.
- As Xcode 11.2 fixes this issue, and is probably a more reliable fix
  than -fno-stack-check, we now require the user update in this case.

Change-Id: I482748fe5f40f82e7daf4da133cdfabbd95ffc89
2019-11-14 09:44:38 +00:00
Sergey Sharybin 54ba6c27b5 Fix missing declaration warnings in Ceres code
This commit includes the following:

- Changes to CMake to make it safer to see which compiler flags are supported,
  so this way we do not need to worry about version checks in CMake.

- Unix platforms (which includes both Linux and Apple as far as i can tell)
  will now enable -Wmissing-declarations warning for the whole Ceres.

- Changes in all sources to solve missing declaration warning. In most cases
  it was either matter of using static qualifier or moving functions to an
  anonymous namespace.

  In one case the function got removed, since it seems to be unused.

  Additionally, in slam examples there was a non-inlined function implementation
  in a header, which is a direct way to cause linking errors if other .cc file
  will include that helper header.

- All third party sources (which is currently only gmock) has this extra
  paranoid warning disabled.

This warning is important in the following cases:

- Detect helper functions which are not needed anymore.
- Avoid unnoticed pollution of namespace.
- Avoid bad level calls.
- Avoid missing updates in header files after changes in implementation file.
- Helps integrating Ceres into software where paranoid warnings are important.

Change-Id: I9b1044aced3910d8c6b2356cfe2bf57f3c8c58db
2019-04-23 12:16:28 +02:00
Alex Stewart ee7e2cb3c8 Set Homebrew paths via HINTS not CMAKE_PREFIX_PATH
- Adding the Homebrew install prefix to CMAKE_PREFIX_PATH, irrespective
  of whether it is a standard location (e.g. /usr/local) overrides any
  user-specified HINTS locations according to the search order of
  CMake's find_xxx() functions.
- Now we append the relevant include/lib directory to its respective
  HINTS directory variable (to the back to avoid overriding any user
  specified values).  This achieves the same effect, but only after
  searching any user specified values.
- Also adds Homebrew install prefix to find_package() calls for
  dependencies optionally exported with CMake after the possible user
  specified <DEPENDENCY_NAME>_DIR prefix.
- Raised as issue #431

Change-Id: I47030f0f4fd9b96665fac57279be2285d9700b9a
2019-03-07 22:12:44 +00:00
Sameer Agarwal 4f8a018531 Revert "Fix custom Eigen on macos (EIGEN_INCLUDE_DIR_HINTS)"
This reverts commit e6c5c7226d.

Reason for revert: <INSERT REASONING HERE>

Change-Id: I38b6a0ae0a7d835d05505d1baae33a52f4752a33
2019-03-07 22:11:03 +00:00
Nikolaus Demmel e6c5c7226d Fix custom Eigen on macos (EIGEN_INCLUDE_DIR_HINTS)
In ceres' CMakeLists.txt a workaround for custom homebrew locations
sets CMAKE_PREFIX_PATH to the homebrew install path. This overrides
explicit requests for custom locations for third-party dependencies,
in particular for Eigen with EIGEN_INCLUDE_DIR_HINTS.

This commit introduces a workaround to make EIGEN_INCLUDE_DIR_HINTS
work on macos by ignoring CMAKE_PREFIX_PATH in that case.

Fixes https://github.com/ceres-solver/ceres-solver/issues/431

Change-Id: I3ec0ec418d45d41b9e5ebdd6aee60552438afec3
2019-03-07 22:08:44 +00:00