On Sat, Mar 26, 2016 at 8:59 AM, Greg Ewing greg.ewing@canterbury.ac.nz wrote:
Also, it would be stretching the string-prefix concept considerably. Currently, the prefixes just represent different ways of specifying a string -- the end result is still always an instance of str. In this proposal, it would be a different type of object with greatly different behaviour.
There's b"..." vs u"...", which do represent entirely different objects, plus f"..." which isn't even a literal at all, but more like a special syntax for an expression (it's more akin to a list display than to a string literal). So there is precedent.
If Path objects had universal support in the stdlib *and* significant support in third-party libraries, they could be the one obvious way to do pretty much anything involving paths. At the moment, they're a cute [1] way of getting just slightly more functionality than strings give. The question is: Should syntax precede or follow extensive usage?
ChrisA [1] Overloading division doesn't really do anything for you, other than the way it looks similar to the normal path sep. It's really more of a concatenation operation, which would normally be + not /.