# Introduction

yotta is a tool that we’re building at mbed to help ourselves and others build better software for C-family languages by making it easier to share and re-use software modules. It’s a command line tool yotta, but also a culture of building software components that do one thing well, declare a clear interface, and can be re-used.

Get started by installing yotta, and following the tutorial. yotta is still early in development, so if you have questions/feedback or issues, please report them on our github issue tracker.

# Installation

yotta is written in python, and distributed using pip, the python package manager. You will need a working installation of both python and pip to install yotta, as well as a working development environment for compiling software, including:

  • CMake, the build system that yotta uses.
  • a compiler, to actually compile the code into working programs.

yotta supports compiling with different compilers by specifying different targets for the compilation, and a compilation target may have its own specific requirements.

To install yotta, please follow the detailed installation instructions for your operating system below:

If you have a complex python setup on your system, you may want to consider installing yotta in a virtualenv, in order to separate its dependencies from other python programs on your system.

To upgrade an existing installation to a new version, see upgrading (the same for all systems).

If you have a restrictive firewall setup, then for yotta to be fully functional you may also need to ensure that it can access the required domains.

# Installing On OS X

# Manual Installation

First install homebrew, a package manager for OS X that we’ll use to install all of yotta’s dependencies.

Next, tap the ARMmbed brew formulae, which lets brew install packages from the mbed team:

brew tap ARMmbed/homebrew-formulae

Now we can install everything that yotta needs:

brew install python cmake ninja arm-none-eabi-gcc

And install yotta itself:

pip install yotta

# Cross-compiling from OS X

To cross-compile, you need the arm-none-eabi-gcc cross-compiler. You can install this using homebrew, after tapping the ARMmbed homebrew package repository.

brew install arm-none-eabi-gcc

To use this compiler, you’ll need to select a supported cross-compilation target, such as frdm-k64f-gcc, by running yotta target frdm-k64f-gcc before building.

# Using Xcode’s compiler to build natively for OS X

To compile things natively you need to have the Xcode command line tools installed. Install Xcode from the Mac app store, then run:

xcode-select --install

To use this compiler to build a module, you should run yotta target x86-osx-native before building. This selects the yotta target description for the native compiler.

# Solving Common OS X installation problems

On OS X, if you get an unknown argument error when running pip install yotta, it means some of yotta’s dependencies have not yet been updated to support Xcode 5.1. To fix this, install yotta by running:

ARCHFLAGS=-Wno-error=unused-command-line-argument-hard-error-in-future pip install yotta

# Installing On Linux

# Installing Dependencies

First install yotta’s dependencies using your system’s package manager. Use whatever 2.7.* python version is provided by your distribution (python 3 support is currently experimental).

on Debian and Ubuntu:

sudo apt-get update && sudo apt-get install python-setuptools  cmake build-essential ninja-build python-dev libffi-dev libssl-dev && sudo easy_install pip

and on Fedora Linux (tested on FC21):

# install development tool dependencies
sudo yum install python-pip cmake ninja-build python-devel libffi-devel openssl-devel clang
sudo yum groupinstall "Development Tools" "Development Libraries"

# update pip to latest release
sudo yum remove python-pip
curl -o get-pip.py https://bootstrap.pypa.io/get-pip.py
sudo python get-pip.py

or under cygwin (on windows), which presents a linux-like environment to yotta:

  • install the windows dependencies
  • install the libffi-developer and openssl-developer cygwin modules as both binary and source
  • install python and pip in cygwin

# Install yotta Itself

After installing the dependencies, install yotta itself using pip (you may need to use sudo for this, depending on your configuration):

pip install yotta

If you experience problems with a systemwide yotta installation on linux, you can try installing yotta in a virtualenv by following these instructions.

You can use the following commands to allow the current user to override module dependencies using yotta link without sudo:

sudo mkdir -p /usr/local/lib/yotta_modules
sudo chown $USER /usr/local/lib/yotta_modules
chmod 755 /usr/local/lib/yotta_modules

# Cross-compiling from Linux

To cross-compile yotta modules for embedded targets, you first need install the arm-none-eabi-gcc compiler.

On most Linux distributions (although not Ubuntu), this can be done by running:

sudo apt-get install gcc-arm-none-eabi

On Ubuntu it’s necessary to use the ARM-maintained gcc-arm-embedded package, instead:

# remove the built-in package, if installed:
sudo apt-get remove binutils-arm-none-eabi gcc-arm-none-eabi
# set up the PPA for the ARM-maintained package:
sudo add-apt-repository ppa:team-gcc-arm-embedded/ppa
sudo apt-get update
# install:
sudo apt-get install gcc-arm-embedded

To use this compiler, you’ll need to select a supported cross-compilation target, such as frdm-k64f-gcc, by running yotta target frdm-k64f-gcc before building.

# Using clang to build natively for Linux

Install a native compiler, such as clang:

sudo apt-get install clang-3.5

To use this compiler to build a module, you should run yotta target x86-linux-native before building. This selects the yotta target description for the native compiler.

# Solving Common Linux installation problems

If you are having trouble with pip not installing yotta, try running sudo pip install -U pip to update your pip installation. Check that your pip installation is up to date by running pip -V, you should get a response of 7.1.2 or greater.

On Ubuntu the default pip installation python-pip is out of date (1.5.2) and cannot upgrade itself via sudo pip install -U pip. To solve this you will need to install pip from easy_install by running easy_install pip. You should then be able to install yotta by running pip2 install yotta.

If you encounter problems with the cryptography library, e.g. 'module' object has no attribute 'X509_up_ref', manually upgrading the cryptography library may fix your issue:

pip install --upgrade cryptography

You can also try installing pip from the Pypy registry if everything else fails.

# Installing on Windows

# Manual Installation

  1. Install python. You must install python 2.7.9 or later for yotta to work on windows. Select either the x86-64 installer if you use 64-bit windows, or the x86 installer if you use 32-bit windows.

    During installation, be sure to select the “add to path” option. This will let you run python easily from a command prompt.

  2. Install CMake. yotta uses CMake to generate makefiles that control the build. Select the latest available version, currently 3.2.1 The 32-bit version will work on all versions of windows. Be sure to check the “add cmake to the path for current user” option during installation.

  3. Install Ninja, the small and extremely fast build system that yotta uses. Download the release archive from the releases page, and extract it to a directory (for example C:\ninja).

  4. Add the directory you installed Ninja in to your path.

  5. Install the arm-none-eabi-gcc cross-compiler in order to build software to run on embedded devices.

  6. Finally, open cmd.exe and run pip install -U yotta to install yotta itself.

# Cross-compiling from Windows

To use yotta to cross-compile binaries to run on embedded hardware, you need to first install the arm-none-eabi-gcc compiler. At the time of writing this, the latest version used for cross-compiling with yotta is gcc 4.9. Download and install it, then add the bin/ subdirectory of the installation directory to your path. After you do that, you should be able to open cmd.exe and run arm-none-eabi-gcc from the command prompt. If that doesn’t work, make sure that your path is properly set.

To use this compiler, you’ll need to select a supported cross-compilation target, such as frdm-k64f-gcc, by running yotta target frdm-k64f-gcc before building.

# Building programs natively to run on windows

yotta does not yet allow compiling programs to run on windows. If you are adventurous and get it working, submit a pull request to update these docs.

# Solving Common Windows Installation Problems

error: command ['ninja'] failed

If you get an error when running yotta build which looks something like this:

':' is not recognized as an internal or external command,
operable program or batch file.
ninja: build stopped: subcommand failed.
error: command ['ninja'] failed

This is caused by re-trying a yotta build after fixing a missing cross-compiler installation. After completing the installation of the compiler, you’ll need to run yotta clean before running build again.

# Adding things to your PATH in windows

Your PATH environment variable holds the location of programs that can be easily executed by other programs. If yotta fails to find one of its dependencies (such as cmake.exe) the first thing to check is that you have added the directory that contains the dependencies executable to the PATH. To add things to your path:

  1. Right click on Computer, select Properties
  2. Select Advanced System Settings
  3. Select the Advanced tab
  4. Click the Environment Variables button
  5. Find the Path variable, edit it, and append the path you want to add, preceded by a semicolon, for example: ;C:\Path\to\wherever

    NOTE: be careful not to add any spaces before or after the semicolon, this can cause commands to fail later.

  6. finally, close then re-open any open cmd.exe windows

# Installing in a Virtualenv

Virtualenv is a way of separating different python programs installed on the same system from each other. If you have a complex python environment on your system it’s recommended that you install yotta inside a virtualenv. To do this, first install the non-python dependencies following the normal instructions for your platform, then:

  1. Ensure you have virtualenv itself installed. It can be installed with:

    pip install virtualenv

    Check that your installation is succesful by running virtualenv --version.

  2. Create a directory to use for the yotta virtualenv:

    mkdir yotta-venv
    virtualenv ./yotta-venv
  3. activate the new virtualenv:

    source ./yotta-venv/bin/activate
  4. install yotta in the virtualenv:

    pip install yotta
  5. (optional) add the binary directory of your virtualenv to your PATH: (if you omit this step, you will need to run yotta as ../path/to/yotta-venv/bin/yotta instead of simply yotta.

    export PATH="/path/to/yotta-venv/bin:$PATH"

Now yotta should work as normal. You will need to activate the virtualenv any time you want to run yotta commands (you can deactivate it afterwards, by simply running deactivate).

# Upgrading yotta (all platforms)

To update yotta itself, run:

pip install -U --no-deps yotta
pip install yotta

This will update yotta to the latest available version, and then install any missing dependencies required by the new version.

You can also run:

pip install -U yotta

This will also attempt to update all of yotta’s dependencies to their latest versions.

On Linux and OS X you may have to run these commands as sudo pip ...., if permission is denied.

# Setting up Tab Completion

yotta uses argcomplete to provide tab completion, so you can set up completion on Linux and OS X by adding the following to your .bashrc or .bash_profile file:

eval "$(register-python-argcomplete yotta)"
eval "$(register-python-argcomplete yt)"

For more detailed instructions, see the argcomplete documentation.