We few. We happy few.
I admit to having strong opinions. As I’ve mentioned in previous blogs, I think Python is the right programing language for Mixed Integer Programming. Python has a consistent, logical design that makes it easy to learn, particularly for people with strong analytical skills (such as the MIP community). It’s made even easier to learn by its high adoption rate, which has lead to a surfeit of educational material as both books and online queries.
Although I’m less pushy about it, I also have strong feelings about what type of computer to use. My choice is MacOSX machines. I prefer using UNIX commands for a lot of the same reasons I prefer Python - it’s well thought and logically consistent (for the most part) and its high adoption rate means Google-hive-mind advice is readily available. I prefer MacOSX over Linux because the GUI makes things easy for other members of my family. (And, I admit, for me as well.)
That said, the combination of Python, MacOSX and MIP does seem to place me amongst a relatively small group of developers. Although each one of these technologies is widely adopted on its own, the Venn diagram of all three seems to be somewhat small. Other than the occasional install-MIP-package-for-OSX headache, this doesn’t create problems. But when those problems do arise, the answers aren’t always easily found.
With that in mind, I’ve created an Install MIP Package on OSX page for the ticdat wiki. If you are one of the ‘happy few’ building MIP engines with Python on OSX, I hope you find it useful.