Merge branch 'main' into feature/dataset-add-tobuffer-labelset-swap

Author: tremblap

README.md

@@ -4,9 +4,16 @@ This library comprises a suite of C++ algorithms used by the Fluid Corpus Manipu
 
 # Dependencies
 
-* [CMake](http://cmake.org) (>=3.11)
+
+* [CMake](http://cmake.org) (>=3.18)
 * [Eigen](https://gitlab.com/libeigen/eigen) (3.4)
 * [HISSTools Library](https://github.com/AlexHarker/HISSTools_Library)
+* [Spectra](https://github.com/yixuan/spectra)
+* [Memory](https://github.com/foonathan/memory.git)
+* [fmt](https://github.com/fmtlib/fmt)
+* [nlohmann JSON](https://github.com/nlohmann/json)
+
+
 
 # Building
 It is unlikely that you will need or want to build this repository directly. Rather, the pattern is to build from one of the repositories for a specific creative coding host, which uses this library as a dependency:
@@ -25,7 +32,7 @@ To build these, clone the repo, change to its directory, and run CMake:
 mkdir -p build && cd build
 cmake ..
 ```
-By default, CMake will download Eigen and the HISSTools library directly at this point. This is the simplest option, and guarantees that the versions used match the versions we build and test against.
+By default, CMake will download the dependencies above directly at this point. This is the simplest option, and guarantees that the versions used match the versions we build and test against.
 
 However, to use versions already on your file system, you can set CMake cache variables `EIGEN_PATH` and `HISS_PATH`:
 
@@ -38,9 +45,9 @@ Because CMake is a system for generating build scripts, rather than a build syst
 * On Windows, Visual Studio can consume CMake projects directly. When used this way, the cache variables are set in a `JSON` file that MSVC uses to configure CMake.
 
 # Portability
-The code base uses standard-compliant C++14 and, as such, should be portable to a range of platforms. So far, it has been successfully deployed to macOS (>= Mac OS X 10.7, using clang); Windows (10 and up, using MSVC); and Linux (Ubuntu 16.04 and up, using GCC), for 32-bit and 64-bit intel architectures. Please check that your compiler version supports the full C++14 feature set.
+The code base uses standard-compliant C++17 and, as such, should be portable to a range of platforms. So far, it has been successfully deployed to macOS (>= Mac OS X 10.9); Windows (10 and up, using MSVC); and Linux (Ubuntu 20.04 and up, using GCC), for 32-bit and 64-bit intel architectures, as well as Apple Silicon. Please check that your compiler version supports the full C++17 feature set.
 
-In principle, it should be possible to build for other architectures, but this has not yet been properly explored and certain parts of the build process may make assumptions about Intel-ness.
+In principle, it should be possible to build for other architectures, but this has not yet been properly explored.
 
 If you find objects causing 'illegal instruction' segmentation faults, it is likely that our default vector instruction flag isn't supported by your CPU.
 
@@ -50,7 +57,9 @@ And *only building for your own machine*? You want to enable the maximal set of
 
 Pass `-DFLUID_ARCH=-mnative` when you run CMake. This tells the compiler to figure out what instruction sets your personal CPU supports, and enable them. This implies a possible performance gain in return for a portability loss.
 
-### More Detail
+
+
+<!--### More Detail
 By default, on Intel / AMD, we enable AVX instructions (`-mavx` on clang/GCC; `/ARCH:AVX` on MSVC). These should work on all relatively recent CPUs (from, say, 2012 onwards). On ARM (with the Bela platform in mind), we use the following architecture flags:
 
 ```
@@ -59,7 +68,7 @@ By default, on Intel / AMD, we enable AVX instructions (`-mavx` on clang/GCC; `/
 
 If your needs are different, then these defaults can be overridden by passing the desired compiler flags to CMake via the `FLUID_ARCH` cache variable.
 
-If your Intel / AMD chip is too old to support AVX, it probably still supports SSE. On macOS and Linux, `sysctl -a | grep cpu.feat` can be used to produce a list of the various features your CPU supports.
+If your Intel / AMD chip is too old to support AVX, it probably still supports SSE. On macOS and Linux, `sysctl -a | grep cpu.feat` can be used to produce a list of the various features your CPU supports.--> 
 
 ## Credits 
 #### FluCoMa core development team (in alphabetical order)

include/flucoma/clients/common/MessageSet.hpp

@@ -189,7 +189,7 @@ MessageSet<std::tuple<Args...>> defineMessages(Args&&... args)
   return {std::forward<Args>(args)...};
 }
 
-auto NoMessages = defineMessages();
+inline auto NoMessages = defineMessages();
 
 } // namespace client
 } // namespace fluid