Re[2]: Re: New stackless proposal

On Sun, 11 Jul 2004 16:51:40 -0700 Paul Prescod <[email protected]> wrote:

> Mark Hahn wrote:
> 
> >...
> >>I notice you use a wrapping model here rather than an inheritance
> >>model (opposite of properties). OSThread looks like a wrapper.
> > 
> > 
> > It is an inheritance model.
> 
> I see nothing in the example to indicate that OSThread inherits from 
> thread1 or vice versa.
> 
> > ... It just happens to reuse the thread1 binding.
> 
> Can you see how that makes it more difficult to read the code?
> 
> >>I don't understand why you use call_ to arm a thread and then rebind
> >>the name. If the call_ function mutates the underlying object then
> >>why do 
> >>you have to rebind the name? 
> > 
> > 
> > That is a mistake.  I will change it to remove the assignment:
> > 
> >      thread1(10)  # call thread1 to "arm" it
> 
> I think using call_ for this is confusing. If you really feel like you 
> have to use call_ it would be a lot more clear to use it for starting. 
> But perhaps better not to use it at all.
> 
> thread1.setParams(10)
> thread1.start()
> 
> Explicit is better than implicit!
> 
call_ is for generator compatibility. It is awkward but allows threads to
be used to construct generators.

# Define function for generator. It is just another Func.
def myrange(n):  # return 0..(n-1)
    for i in n:
        yield(i) # just another function call

# Make myrange a thread. Now it can be a generator (coroutine).
generator = Thread(generator)

# Arm myrange with argument 5 and call its iter_ method to create an
# iterator that coordinates with the generator's yield() function
# calls.
for x in myrange(5):
    print(x)

Then again, with decorators:

(Thread)
def myrange(n):
    ...

Lenard Lindstrom
<[email protected]>