rig_c_sa: Python C module for the Rig simulated annealing placer C kernel

This Python package contains an efficient set of datastructures and manipulation functions for implementing a kernel for Rig's simulated annealing placement algorithm with a wafer-thin CFFI based interface.

The library is not intended for standalone usage and, as a result, the whole API should be considered "internal" and unstable for public use.

Running tests locally

The C library is tested using using the check library. The test suite can be built using the following command:

$ gcc -std=c99 -g -o run_tests -Irig_c_sa tests/*.c rig_c_sa/sa.c -lm $(pkg-config --cflags --libs check)

The test suite should then be run under valgrind to ensure any memory leaks are found:

$ valgrind -q --leak-check=full ./run_tests

Continuous Integration and Deployment to PyPI

When committing changeds to GitHub tests (and/or compilation checks) are run against:

• Travis CI will test the code on Linux and Mac OS X.
• AppVeyor will test the code on Windows 32- and 64-bit.

A PyPI release is automatically created when a new tag is pushed to GitHub. To create a new rig_c_sa release, please follow these steps:

• Increment the version number in rig_c_sa/version.py according to Semantic Versioning 2.0.0 in master.
• Ensure that ensure that all tests on Travis and AppVeyor pass (critical!)
• Create a new tag (git tag -a vX.Y.Z) which describes the changes since the last release.
• Push this to GitHub. Travis and AppVeyor will automatically build and deploy the tagged release to PyPI. This deployment includes:
  • Source distribution
  • Precompiled OS X binaries
  • Precompiled Windows 32- and 64-bit binaries

The ugly workings of this process are described more fully in .travis.yml and appveyor.yml.