Bjoern Schliessmann a écrit : > Bruno Desthuilliers wrote: > >>Bjoern Schliessmann a écrit : > > >>>You can't just declare in Python, you always define objects (and >>>bind a name to them). >> >>def toto(): >> global p >> p = 42 >> >>Here I declared 'x' as global without defining it. > > > Ah well, someone had to notice it ... > > BTW, where's x? :)
Sorry. s/x/p/g, of course. > >>>Yes, globals need to be defined before you >>>can access them using "global". >> >>For which definition of 'defined' ? > > > to define a name: to bind an object to a name Then - according to this definition - your assertion is false. > >> >>> p >>Traceback (most recent call last): >> File "<stdin>", line 1, in <module> >>NameError: name 'p' is not defined >> >>> toto() >> >>> p >>42 >> >>> > > > Easy, p = 42 is a definition. Yes, but it came *after* the global statement. So the global statement *is* a kind of declaration, and the 'subject' of this declaration doesn't need to be previously defined. > >>But anyway: in Python, everything's an object, so the only >>thing that makes functions a bit specials is that they are >>callable - as are classes, methods, and every instance of a class >>implementing __call__ FWIW. So saying 'variables holds data, >>functions do stuff' is unapplyiable to Python. > > > I don't think so. Even if everything is an object, there is still a > concept behind the words "function" and "variable". f = lambda x: x+2 def toto(c): return c(21) toto(f) What's f ? A function or a variable ?-) -- http://mail.python.org/mailman/listinfo/python-list
