The vale of tears that is Mavericks

So I thought to myself: sure, it’s an Apple update, why not change my machines to 10.9 (“Mavericks”)? This was my entry into the system-administration-hell that is OSX Mavericks. The appeal of OSX for science users was always that it’s a supported hardware platform with the few paid applications that you really need (mainly Mathematica and Papers, and every so often something weird like Final Cut or Photoshop) running on top of a functional and programmable Linux box. Nobody ever thought Darwin was a particularly¬†good linux distribution. But it basically worked, and you could get work done.

And then along came Mavericks. In which everything that you learned to live with in the previous Darwins has simply been savagely trashed and gratuitously broken.

Where to start with this? They removed the debugger. The debugger. Yes, Apple, many of us still use the gcc/gdb toolchain because XCode (while it’s great for making gee-whiz windowing applications in Objective-C) offers essentially nothing to the scientific programmer except a complete lack of portability. And icons. Lots and lots of icons. And menus. But don’t worry! They replaced gdb with the (experimental) lldb. A debugger which doesn’t even have an emacs mode associated with it.

Polymake doesn’t work. I have an entire extra system that I’m keeping alive just because it runs polymake.

Valgrind support is “experimental and mostly broken”. Valgrind!

It’s taken me weeks just to port my old research code over to the new machines, and the job is still not really done.¬†I don’t even want to go into the various graphics issues. Does CUDA work? OpenCL? Color management? This thing is the Windows Vista of Mac operating systems. I’m seriously thinking about scrapping the whole enterprise and going to Ubuntu.






Comments are Disabled