Indeed, I'm a scientific programmer, and I've noticed that when most scientists talk about "Python," they don't make a clear distinction between where the language ends and the libraries begin. If they even know. Most users, myself included, download some big installer like Anaconda or WinPython, and off we go.
Same. Obviously a OO scripting language by itself isn't fast enough for most scientific coding, but when the C/Fortran libraries are tightly integrated with lots of stats/plots and other libraries with great IDE/database...etc it makes for a great and free/open scientific modeling platform. It's really the whole package and the language is just 1 component of that.
Indeed, and it's my understanding that supporting that kind of integration is a strength of Python. Since I use Python in the lab, I've also found that the whole ctypes thing is a life saver when dealing with hardware drivers that are only furnished with a C API.
