LBRYCommunity/kivy-ios
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity
kivy-ios/README.md
Richard Larkin 06e6e0b8b5
Document lack of multiprocess / async functionality (#402) 
* docs: Added note regarding iOS multiprocess/tghreading limitations

* Fix typo

* Tweaked wording

* 🔧 Refine accuracy of multiprocess docs

* ✨ Fix typo
2019-11-07 21:46:54 +02:00

11 KiB
Raw Blame History

Kivy for iOS

Backers on Open Collective Sponsors on Open Collective

This toolchain is designed to compile the necessary libraries for iOS to run your application and manage the creation of the Xcode project.

We do not provide any binary distributions of this toolchain. You do need to compile it at least once before creating your Xcode project.

The toolchain supports:

  • iPhone Simulator (x86_64)
  • iPhone / iOS (armv7 and arm64)

You can select between Python 2.7 or Python 3.7 by specifying the recipes python2 or python3 when building.

These recipes are not ported to the new toolchain yet:

  • lxml

Requirements

Currently, the toolchain requires a few tools for compilation. You will need:

  • Ensure you have python3 installed - this is needed for toolchain.py:

    brew install python

  • Ensure you have the right dependencies installed for python3:

    pip3 install -r requirements.txt

  • Xcode 10 or above, with an iOS SDK and command line tools installed:

    xcode-select --install

  • Using brew, you can install the following dependencies:

    brew install autoconf automake libtool pkg-config
brew link libtool

  • Install Cython (0.28.1):

    # pip method if available (sudo might be needed.)
pip3 install cython==0.28.1

Using the toolchain

Any Python extensions or C/C++ library must be compiled: you need to have what we call a recipe to compile it. For example, Python, libffi, SDL2, SDL_image, freetype... all the dependencies, compilation and packaging instructions are contained in a recipe.

You can list the available recipes and their versions with:

$ python3 toolchain.py recipes
audiostream  master
click        master
cymunk       master
distribute   0.7.3
ffmpeg       2.6.3
ffpyplayer   v3.2
flask        master
freetype     2.5.5
hostlibffi   3.2.1
hostpython2  2.7.1
hostpython3  3.7.1
ios          master
itsdangerous master
jinja2       master
kivy         1.10.1
libffi       3.2.1
libjpeg      v9a
libpng       1.6.26
markupsafe   master
moodstocks   4.1.5
numpy        1.16.4
openssl      1.0.2k
photolibrary master
pil          2.8.2
pillow       6.1.0
plyer        master
pycrypto     2.6.1
pykka        1.2.1
pyobjus      master
python2      2.7.1
python3      3.7.1
pyyaml       3.11
sdl2         2.0.8
sdl2_image   2.0.0
sdl2_mixer   2.0.0
sdl2_ttf     2.0.12
werkzeug     master

Then, start the compilation with:

$ python3 toolchain.py build python3 kivy

You can build recipes at the same time by adding them as parameters:

$ python3 toolchain.py build python3 openssl kivy

Recipe builds can be removed via the clean command e.g.:

$ python3 toolchain.py clean openssl

You can install package that don't require compilation with pip::

$ python3 toolchain.py pip install plyer

The Kivy recipe depends on several others, like the sdl* and python recipes. These may in turn depend on others e.g. sdl2_ttf depends on freetype, etc. You can think of it as follows: the kivy recipe will compile everything necessary for a minimal working version of Kivy.

Don't grab a coffee, just do diner. Compiling all the libraries for the first time, 3x over (remember, 3 archs, x86_64, armv7, arm64) will take time.

For a complete list of available commands, type:

$ python3 toolchain.py

Create the Xcode project

The toolchain.py can create the initial Xcode project for you::

$ python3 toolchain.py create <title> <app_directory>
$ python3 toolchain.py create Touchtracer ~/code/kivy/examples/demo/touchtracer

Your app directory must contain a main.py. A directory named <title>-ios will be created, with an Xcode project in it. You can open the Xcode project using::

$ open touchtracer-ios/touchtracer.xcodeproj

Then click on Play, and enjoy.

Did you know ?

Everytime you press Play, your application directory will be synced to the <title>-ios/YourApp directory. Don't make changes in the -ios directory directly.

Configuring your App

You can configure and customize your app in various ways:

  • Set the icon and launch images in XCode. Note that XCode requires that you specify these assests per device or/and iOS version.

  • When you first build your XCode project, a 'main.m' file is created in your XCode project folder. This file configures your environment variables and controls your application startup. You can edit this file to customize your launch environment.

  • Kivy uses SDL, and as soon as the application starts the SDL main, the launch image will disappear. To prevent that, you need to have 2 files named Default.png and Default-Landscape.png, and put them in the Resources folder in Xcode (not in your application folder)

Did you know ?

If you wish to restrict your apps orientation, you should do this via the 'export_orientation' function in 'main.m'. The XCode orientation settings should be set to support all.

Using recipes

Recipes are used to install and compile any libraries you may need to use. These recipes follow the same format as those used by the Python-for-Android sister project. Please refer to the recipe documentation there for more detail.

Reducing the application size

If you would like to reduce the size of your distributed app, there are a few things you can do to achieve this:

  • Minimize the build/pythonX/lib/pythonXX.zip: this contains all the python modules. You can edit the zip file and remove all the files you'll not use (reduce encodings, remove xml, email...)

  • Go to the settings panel > build, search for "strip" options, and triple-check that they are all set to NO. Stripping does not work with Python dynamic modules and will remove needed symbols.

  • By default, the iOS package compiles binaries for all processor architectures, namely x86_64, armv7 and arm64 as per the guidelines from Apple. You can reduce the size of your ipa significantly by removing the x86_64 architecture as they are used only for the emulator.

    The procedure is to first compile/build all the host recipes as is:

    python3 toolchain.py build hostpython3

    Then build all the rest of the recipes using --arch=armv7 --arch=arm64 arguments as follows:

    python3 toolchain.py build python3 kivy --arch=armv7 --arch=arm64

    Note that these packages will not run in the iOS emulators, so use them only for deployment.

Usage

python3 toolchain.py <command> [<args>]

Available commands:
    build         Build a recipe (compile a library for the required target
                    architecture)
    clean         Clean the build of the specified recipe
    distclean     Clean the build and the result
    recipes       List all the available recipes
    status        List all the recipes and their build status

Xcode:
    create        Create a new xcode project
    update        Update an existing xcode project (frameworks, libraries..)
    launchimage   Create Launch images for your xcode project
    icon          Create Icons for your xcode project
    pip           Install a pip dependency into the distribution
    pip3          Install a pip dependency into the python 3 distribution

FAQ

Fatal error: "stdio.h" file not found

You need to install the Command line tools: xcode-select --install

You must build with bitcode disabled (Xcode setting ENABLE_BITCODE should be No).

We don't support bitcode. You need to go to the project setting, and disable bitcode.

You don't have permissions to run

It is due to invalid archs, search for them and check it. Maybe you targetted a simulator but have only armv7/arm64. Maybe you want to target your iPad but it as only x86_64.

Why does the python multiprocess/subprocess module not work?

The iOS application model does not currently support multi-processing in a cross-platform compatible way. The application design focuses on minimizing processor usage (to minimize power consumption) and promotes an alternative concurrency model.

If you need to make use of multiple processes, you should consider using PyObjus to leverage native iOS functionals for this.

Support

If you need assistance, you can ask for help on our mailing list:

We also have a Discord channel:

Contributing

We love pull requests and discussing novel ideas. Check out our contribution guide and feel free to improve Kivy for iOS.

The following mailing list and IRC channel are used exclusively for discussions about developing the Kivy framework and its sister projects:

Discord channel:

License

Kivy for iOS is released under the terms of the MIT License. Please refer to the LICENSE file.

Backers

Thank you to all our backers! 🙏 [Become a backer]

Sponsors

Support this project by becoming a sponsor. Your logo will show up here with a link to your website. [Become a sponsor]