(Assuming (use mpi s11n))
(continuation-capture (lambda (x) (MPI:send (serialize `(process ,x)) dest tag
comm)))
Where:
* x: the current continuation
* `(process DATUM): a message in a hypothetical high-level protocol
for process migration
* dest: the destination process ID (integ
On Tue, May 20, 2008 at 5:22 PM, Ivan Raikov <[EMAIL PROTECTED]> wrote:
> Well, of course, it is "possible" to implement anything in a
> Turing-equivalent language, but I am curious what features of Erlang
> or Termite are difficult to implement with MPI primitives and
> Scheme. For example, "proc
Answers embedded below:
Alaric Snell-Pym <[EMAIL PROTECTED]> writes:
> Well, as far as I can tell, MPI seems to have a fairly static view of
> the world as having a fixed number of processes in it upon startup,
Correct. This helps with communication transport optimizations --
when you create
Ok, point taken about dynamic spawning of new processes. We have hot
code loading and modules in Chicken. MPI has MPI_Disconnect,MPI_Wait
and MPI_Test, which are not incorporated in the mpi egg, but could be
used for monitoring process termination. The group communication
facilities of MPI are u
On 20 May 2008, at 3:27 pm, Ivan Raikov wrote:
Perhaps you
would like to show some code examples? I don't know who is this
"process-management model" and why I care about it, but I would like
to see some examples of actual working code that cannot be reproduced
with the MPI library.
Well, as
Well, of course, it is "possible" to implement anything in a
Turing-equivalent language, but I am curious what features of Erlang
or Termite are difficult to implement with MPI primitives and
Scheme. For example, "process migration" in Scheme MPI could be simply
passing a continuation from one n
On 20 May 2008, at 3:16 pm, Alaric Snell-Pym wrote:
IIRC there's more to it than that. As well as passing messages back
and forth, Erlang also has mechanisms for passing uncaught exceptions
in the top-levels of processes back to processes waiting for messages
from that process... or something l
2008/5/20 Ivan Raikov <[EMAIL PROTECTED]>:
>
>
> Ok, can you show a Termite example that cannot be implemented with MPI?
I'm sure its possble to implement most if not all, but, termite
provides those parts of the erlang model which make the
message-passing model interesting already, without needi
I am confused -- you talk about passing uncaught exceptions back to
processes waiting for messages, and then you talk about preventing
deadlock situations in a seemingly different scenario. Perhaps you
would like to show some code examples? I don't know who is this
"process-management model" and
Ok, can you show a Termite example that cannot be implemented with MPI?
"Leonardo Valeri Manera" <[EMAIL PROTECTED]> writes:
> MPI is very nice Ivan - its not really the same kind of thing as Termite
> though.
>
> In any case, I can't use MPI in any of my projects - its straight GPL.
>
> T
2008/5/20 Ivan Raikov <[EMAIL PROTECTED]>:
>
> There is already "Erlang-style concurrency" in Chicken via the MPI
> egg.
MPI is very nice Ivan - its not really the same kind of thing as Termite though.
In any case, I can't use MPI in any of my projects - its straight GPL.
Termite is BSD, so I'd
On 20 May 2008, at 3:10 pm, Ivan Raikov wrote:
There is already "Erlang-style concurrency" in Chicken via the MPI
egg.
IIRC there's more to it than that. As well as passing messages back
and forth, Erlang also has mechanisms for passing uncaught exceptions
in the top-levels of processes b
There is already "Erlang-style concurrency" in Chicken via the MPI
egg.
-Ivan
"Leonardo Valeri Manera" <[EMAIL PROTECTED]> writes:
> 2008/5/19 Mikael More <[EMAIL PROTECTED]>:
>> Hi!
>>
>> I just emailed with Guillaume Germain, the author of Termite. He said it
>> should be quite easy t
2008/5/19 Mikael More <[EMAIL PROTECTED]>:
> Hi!
>
> I just emailed with Guillaume Germain, the author of Termite. He said it
> should be quite easy to port Termite to Chicken. Is anyone up for that?
Not me, though I'd love to see termite on chicken - erlang-style
concurrency is win :)
Cheers,
Le
On Fri, May 16, 2008 at 12:34 PM, Hans Bulfone <[EMAIL PROTECTED]> wrote:
>
> my question is: will it still be possible to write a (er-) macro that
> creates new identifiers like define-record does? from previous
> postings i got differing impressions.
> i know it's not considered good style to do
On Mon, May 19, 2008 at 11:24 PM, Lui Fungsin <[EMAIL PROTECTED]> wrote:
> So I confirmed that the sigpipe hanging my chicken program stems from
> the same problem - that the signal handler is not re entrant.
>
> If I handle sigpipe, linux will send the signal when the pipe is
> broken. Moreover, s
16 matches
Mail list logo