abusing exceptions for continuations

[gangesmaster]

i've had this strange idea of using the exception's traceback (which
holds the stack frame) to enable functional continuations, meaning,
raise some special exception which will be caught by a reactor/
scheduler/framework, which could later revive it by restoring the
frame.

i'm thinking of using the generator's implementation (some minimal
support on the c-side)

has this been tried before? what were the results?
thanks,
-tomer

[Bruno Desthuilliers]

gangesmaster a écrit :

i've had this strange idea of using the exception's traceback (which
holds the stack frame) to enable functional continuations, meaning,
raise some special exception which will be caught by a reactor/
scheduler/framework, which could later revive it by restoring the
frame.

i'm thinking of using the generator's implementation (some minimal
support on the c-side)

has this been tried before? what were the results?

I don't know, but you may want to have a look at Stackless Python.

[Paul McGuire]

On Dec 10, 2:39 am, gangesmaster <[e-mail]:

i've had this strange idea of using the exception's traceback (which
holds the stack frame) to enable functional continuations, meaning,
raise some special exception which will be caught by a reactor/
scheduler/framework, which could later revive it by restoring the
frame.

i'm thinking of using the generator's implementation (some minimal
support on the c-side)

has this been tried before? what were the results?

thanks,
-tomer

Simpy has a mechanism like this, for discrete event simulation
modeling of processes and resources. Processes "run" for a while,
then yield a delay or a hold on a resource, and are resumed later.

-- Paul

[Lautaro Pecile]

On Dec 10, 5:39 am, gangesmaster <[e-mail]:

i've had this strange idea of using the exception's traceback (which
holds the stack frame) to enable functional continuations, meaning,
raise some special exception which will be caught by a reactor/
scheduler/framework, which could later revive it by restoring the
frame.

i'm thinking of using the generator's implementation (some minimal
support on the c-side)

has this been tried before? what were the results?

thanks,
-tomer

That idea has been tried here http://aspn.activestate.com/ASPN/Coo.../Recipe/474088

It fails with algorithms which not use tail call recursion, and is
veeery slow. Try with Stackless.

Lautaro