Stackless Python Question(s)

Courageous jkraska1 at san.rr.com
Mon May 29 04:25:43 CEST 2000


> I just read a little of what Stackless Python is...and though I do not
> pretend to understand it "all" it looks to be a good thing.

It's a completely wonderful thing as soon as you fall into
an application area which truly needs it.

> What are they good points and bad points about going stackless?

The point of stacklessness is to allow rapid frameshifts and
(virtually) arbitrarily deep recursion. The first class continuation,
which is a snapshot of an execution frame at an arbitrary moment
in time, is one obvious functionality which stacklessness enables.

Continuations themselves allow programmers to build their own
schedulers with incredible ease; this is allowing me to build
a rapid prototyping simulation environment in python that just
does incredible amounts of stuff for a very small linecount.
Instead of timeslicing over actual time, I timeslice over sim-
ulation primitives. Until you need exactly this, it can be hard to
imagine how incredibly wonderful this is.

> Will Python go that way totally?

Well, only Guido can say, really. One snag is that I believe
that it is impossible to implement stacklessness in JPython without
altering the JVM. Plainly this means that stacklessness can't
be done in Pure Java unless Sun, et al endorses it.

C/



More information about the Python-list mailing list