This repository has been archived by the owner on Dec 21, 2023. It is now read-only.
[WIP] Proof-of-concept integration of pure swift model into TuriCreate #2878
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
added 6 commits
December 21, 2019 16:26
This class proposes a new method for managing extensions. It is meant to replace the bulky and unweildy macro system in use currently. The new code is contained in model_server_v2.
Rearchitecture of the the build system. A vast simplification. Changes - ./configure can produce an XCode project of the Turi source code. - deps/ can be compiled and installed out of the regular build process. Helpful for the XCode project. - All packages in src/external/ and src/visualization compile into static libraries. - The rest of the code base compiles into a single shared library. This means that most everything occurs in one single CMakeLists.txt file using standard cmake commands. - All headers are installed into targets/include when `make install` is called. The main library is installed into targets/lib. - The code definitions that reflect compiler oddities (e.g. is std::hash<int128_t> defined) are dumped into a single header file, src/turi_common.h. All source files and headers include this file first. This allows other programs to link against this library / headers reliably. - The python part of TuriCreate now builds through the standard setup.py method. Cython / pybind11 extensions are compiled and packaged using the standard extensions mechanisms and linking against the installed Turi headers. WIP TODO: the python installation, deployment.
This adds an XCode project that demonstrates how to effectively write a swift ML model while allowing it to be called from C++ or python. The key idea is to expose the swift class to objective-C (add @objc modifier), then write a simple Objective-C++ class that is registered using the new model registration mechanism. The type convesion is handled by overloading the variant_type converter to handle native Objective-C types. See src/xcode_swift_toolkit/TC_Swift_Demo/TC_Swift_Demo.xcodeproj for a proof of concept.
hoytak
requested review from
ylow,
TobyRoseman,
znation,
guihao-liang and
nickjong
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR adds an XCode project that demonstrates how to effectively write a swift ML model while allowing it to be called from C++ or python.