[Stackless] Why does xsdb use stackless?

[Aaron Watters]

Well, you can always use an event loop, but this argument reminds
me of a fortran programmer who says that recursion is
not important because you can always emulate it using
a stack.

Unrolling the application at each point where you need to wait
simply turns the code inside out (just like unrolling
a set of recursions).  Threading is the only
acceptable alternative...

...Unless I simply make the concurrency restrictions
harsher and roll back any transaction which needs
to wait

regarding generators: I don't get it, they seem extremely
limited to me.

[I'm cc the stackless list on this for discussion, hope
you don't mind.]

    -- Aaron Watters

Itamar Shtull-Trauring wrote:

>On Wed, 2003-12-10 at 06:09, Aaron Watters wrote:
>
>>Using tasklets to implement db concurrency control is particularly
>>elegant (and much more efficient than using threads, which is the only
>>really acceptable alternative).
>>
>
>Yeah, threads are not the way to go. But an event loop is pretty much
>what the tasklets are running on top of anyway, no? Tasklets can make
>the code cleaner, but I don't think they'd make it fundamentally more
>efficient. That is, "read from channel" is essentially equivalent to
>finishing a function and then having a callback continue processing
>later. I suppose there's overhead of extra Python function calls and
>attribute lookups, solvable in some cases by generators...
>
>Still, it is a shame stackless' functionality never made it into Python.
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.stackless.com/pipermail/stackless/attachments/20031210/3289aa12/attachment.htm>