Hi, Dirk!
I have some questions on project goals.
1. Are you going to create something portable or SBCL/something else only?
2. Are you going to rely on some communication library such as PVM/MPI or something completely independent?
3. Do you have some plan or just experimenting / imitating Erlang?
P.S. I am interested in parallel computation in Common Lisp. I have created UFFI bindings for PVM library. Though I do not remember exact URL of the project :)
Hi Ivan,
Ivan Boldyrev wrote:
I have some questions on project goals.
I've written quite a bit about Erlisp's goals in this roadmap: http://dirkgerrits.com/programming/erlisp/roadmap/ It's sort of the Erlisp manifesto. ;)
- Are you going to create something portable or SBCL/something else only?
Something portable, definately. I try to abstract away from all implementation specifics (see src/compatibility.lisp). Currently, only SBCL specifics are covered, but I do want to expand this in the future. (Patches are always welcome. ;))
- Are you going to rely on some communication library such as PVM/MPI or something completely independent?
At the moment there is no need, as I'm "just" making an Erlang-clone. But when the time comes to go "beyond Erlang" (and it will), I may consider it.
- Do you have some plan or just experimenting / imitating Erlang?
I do have a vague plan (see the roadmap), but mostly I'm just learning on the way. Imitating Erlang is the first step on the long road ahead.
P.S. I am interested in parallel computation in Common Lisp. I have created UFFI bindings for PVM library. Though I do not remember exact URL of the project :)
Is it http://lispnik.newmail.ru/lpvm/ ? I haven't actually used PVM (and have only dabbled in MPI) but I'll try to take a look at it.
Kind regards,
Dirk Gerrits
On 9150 day of my life Dirk Gerrits wrote:
P.S. I am interested in parallel computation in Common Lisp. I have created UFFI bindings for PVM library. Though I do not remember exact URL of the project :)
Is it http://lispnik.newmail.ru/lpvm/ ?
Yep.
I haven't actually used PVM (and have only dabbled in MPI) but I'll try to take a look at it.
PVM is older than MPI; key differences: PVM has simplier API, heterogeneousity and some degree of fault tolerancy.
Perhaps it is better to build own library. Though PVM has at least two advantages: 1. We don't have to implement everything from scratch. 2. There are versions of PVM for specific communication hardware (e.g. Myrinet or ATM).