Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Dabaghi & Der Kiureghian Model, Conan Packaging (#7)
* Busy setting up Windows config for CI * Trigger on windows branch * See if conda works * Try downloading and installing conda using wget * Try again to download conda installer * See if conda is installing * Add conda to path * See where conda is * Using where to find conda... * Added scripts folder to path for conda * See where includes and libs are * Should have environment variables for compilation now * Updated to see how directories look * Yet another try * See why it can't find libs * Most likely IPP will fail, but lets see * Updated cmake files to search in different locations when Windows * CMake should be able to find IPP and MKL now * Testing debug and release on Windows * Change conda installer config * Changed conda to 32-bit to match Pipelines * Try VS2019 * Trying 64-bit conda on VS2017 again * Try to set python to 64 bit * Try specifying 64-bit build using cmake * Specify generator for CMake * Pass config to ctest so it finds appropriate unit tests * Run ctest with verbose flag to see why it is failing * Break-up builds and set environment variables using setx instead of set * Try installing conda for all users * Set variables using set instead of setx * Change build to trigger on master * Busy updating README * Added tested systems to README * Small change * Updated README file * Updated title in Doxygen * Update to make emdash work * Another try for em dash * Fixed typo * Updated gitignore * Started adding model from NOWS for wind loading; need to check methods for going from velocity to pressure before proceeding * Updates to constructor for Wittig model * Added function for calculating wind profile based on exposure category * Overloaded constructor for WittigSinha to handle vector inputs for heights and x, y locations * In process of writing functionality to generate complex random numbers based on cross-spectral density matrix * Busy adding functionality to calculate discrete time series, but first need to add wrapper for inverse FFT using MKL * Added functionality to compute inverse FFT * Added testing for inverse FFT * Added functionality to generate wind velocity profiles at all nodes requested. Still need to write to JSON and unit test most everything for wind * Added functionality to write to JSON, now need to add unit tests * Fixing compiler errors, then will begin adding unit tests * All compiler errors fixed, will add unit tests next * Added unit tests for velocity profile * Added unit testing for cross spectral density matrix computations * Fixed errors in indexing when calculating location time history * Added unit tests for time history generation when floors are all the same height * Added unit tests to check various constructors for WittigSinha, check that specifying same seed value provides identical results * Added more unit tests for WittigSinha generate capability * Changed constructor input units to be in imperial and default outputs to be in m/s * Added wind velocity for velocity profile to pattern in JSON output * Slight change to unit tests for WittigSinha for validation * Changed filter function implementation to avoid having IPP includes in header file * Setting shared and static builds * Static and shared builds working on Linux, will get MacOS and Windows built next * Temporarily trigger Travis build on conan branch * Temporarily trigger Azure build on conan branch * Updated CMake functions to avoid duplication * Updated static build for MacOS * Added static libs for MKL and IPP to Conda * Adding option flags to build script for CMake * Changed MKL and IPP library locations for MacOS * Configuration change for Travis build on Linux with Clang * Changed library suffix for MacOS * Added Intel OpenMP installation through Conda for MacOS * Working on static build for CI * Fixed linking issue related to OpenMP and Clang * Added libiomp to LD_LIBRARY_PATH * Trying to find out why Clang can't find libiomp * Another try with TBB * Another test with TBB... * Trying sequential build * Removed as-needed flag * Changed linker order for MacOS * Forgot to make changes for MacOS linker in helper functions in cmake * Added flag for symbol visibility settings on MacOS * Added visibility flag for library compilation as well * Changed visibility flag to hidden * Fixed error in Windows include dirs for MKL and IPP * Missed one default visibility settting * Configured for all Linux, MacOS and Windows; testing to see if builds succeed on all * Removed some linker flags for MacOS * Fixed issue with flags * Fix for conda on Azure * Another Azure fix attempt * Deactivate conda environment * Adding conda to path * Use conda path to specify MKL and IPP locations * Trying to access conda environment variables * Add conda path as variable * Yet another try at accessing the environment variables * Trying another way... * Trying to access environment variables using batch script syntax * Changed build to use conan to pull in dependencies * Updates to package smelt * Update Travis to build on stable branch * Build on all branches * Trying to trigger build * Update to config to build on stable branch * Updates to config * Another update to config in how environment variable are declared * Troubleshooting config * Config stuff * Fixed whitespace and tab errors in config file * Another config minor change * Fix for branch regular expression config * Checking environment variable * Changed how environment variables are evaluated * Messing with conditionals * Printing env var * Still figuring out conditionals * Changed conditional evaluation of env vars in travis config * Don't think this is going to work... * Try this one... * Set env vars in entire matrix * Added parens to strings for env vars * Removed curly braces * Config testing * Not sure why this isn't working * Added space between square brackets * Update CmakeLists include path for MKL and IPP; changed how pip installs specifying user * Updated CMakeLists to pull in headers for MKL and IPP from Conan * Fixed typo in Travis config * Fixed another typo * See if just packaging works * Restrict builds to only x86_64 * Testing build without conan docker * Trying updated build that specifies arch and build types * Fixed typo in config * Updated conanfile to only build shared or static based on current build--this was causing issues with missing libraries * Another try * Changed linker flags for static compilation * Try Linux without Docker * Try limiting build compiler versions * Try explicitly setting cmake version to conan detected compiler * Yet another try to get this to work... * Add default profile * Added build-essential installation * Removed specification of compiler in builder * Update compiler settings in build.py and add sudo for linux machines * Added build-essential to packages * Attempt to fix issue with linker not finding libs * Changed how linker flags are passed * Minor change to CMakeLists.txt * Changed how link flags are passed for Clang * Removed disable-new-dtags flag * Added unit tests to build * Updated conanfile to see if this fixed upload issue * Added build dir for tests in cmake in conanfile * Another try to find tests * Try calling ctest in different way * Added imports method to make shared libraries available for unit tests * Forgot to pass self as argument to imports * Changed imports to place all shared libs in bin folder * Added shared libs to path in dependencies and then also add shared libs to path for smelt * Updated config to match version of branch * Updated travis config and also conanfile, still need to figure out how make dynamic libraries visible to unit tests * Fixed typo in config and trying to copy dylibs to bin folder for unit tests * Updated conanfile to run copy command for MacOS to put dylibs in bin directory * Fixed error in Travis config * Changed conditionals in scripts * Updated IPP library names for static Windows build * Updated conanfile calls to ctest for Windows to specify either Release or Debug config * Fixed syntax error in conanfily * Added smelt to library path for MacOS unit tests; fixed another typo related to Windows builds * Update to how shared unit tests are called on MacOS * Fixed typo in conanfile * Changed build config for shared windows * Fixed location of libs for Windows package * Changed conditional for build type on Release * Add lib folder to path for Windows * Still trying to figure out shared run failure on Windows * Added package info for Windows * Updating conanfile to get sources from git * Updated linking for Windows unit tests * Trying to build again after changing how Intel libraries are packaged * Try build with updated MKL & IPP shared conan packages * Changed build policy * Changed default setting for shared to False * Changed build policy back to previous config--build missing * Updated conanfile and CMakeLists.txt to pull in all external dependencies through conan * Deleted all external dependency sources since managed through conan * Changed include calls to match conan package structure * Removed print message from CMakeLists.txt * Updated Travis and Appveyor configs with CONAN_REMOTES * Fixed syntax errors in YAML configs * Try to force boost to build from source * Add includedir to package info * Updated recipe so that headers are correctly copied * Updated conanfile to include template files in includes and pull in shared or static libraries based on build * Fixed typo in conanfile * Updated README links * Update README.md * Started adding near-fault model by Dagaghi & Der Kiureghian * Added functionality to calculate number of pulse motions * Added regression constants * Busy adding function to transform parameters from normal to original space * Finished adding functionality to transform parameters to real space from normal space * Busy adding functionality to do Nelder-Mead optimization of parameter calcs * Added Nelder-Mead optimization, but still need to review code to make sure it makes sense * Finished adding functionality to backcalculate modulating parameters * Busy adding function to simulate modulated filtered white noise * Added functionality for generating modulated filtered white noise * Moved acausal filter to filter.h, added functionality to scale accelerations based on Arias intensity * Added functionality to include pulse acceleration in time history * Added functionality to truncate time histories * Added functionality to baseline correct accelerations; added numeric utility function for polynomial fitting, derivatives and evaluation * Added functionality to generate time histories; now it all needs to be tested * Unit tested all functions added to numeric utils * Added unit tests for uniform distribution * Nelder-Mead optimization unit testing completed * Busy fixing compiler errors in DabaghiDerKiureghian implementation * Compilation errors fixed in DababhiDerKiureghian, will unit test next * Busy adding unit tests * Still unit testing, finding and fixing bugs * Fix to white noise generation * Unit tested acuasal HP filter * Fixed block operations in ground motion simulation * More unit tests, just need to check generation functionality then can do integration testing * Unit testing completed, will validate next * Fixed bug in pulse acceleration calculations * Fixed version number in conanfile.py
- Loading branch information
1 parent
7f6cfa0
commit 6188227