[racket] Remote execution in Racket

From: Jay McCarthy (jay.mccarthy at gmail.com)
Date: Fri Aug 20 10:20:40 EDT 2010

There's no way to dump or load images in Racket. You'll have to
implement state dumping/loading in your application.

Jay

On Fri, Aug 20, 2010 at 2:52 AM, Frederick Ross <madhadron at gmail.com> wrote:
> I find myself with a rather peculiar problem, and I'd love anyone
> else's input on it.
>
> I'm writing a piece of software for our core facility.  Among the
> things it does is wrap up a bunch of unix commands into one atomic
> unit, handle their inputs and outputs, and keep track of all the files
> and commands involved in an SQLite database.  The thing about
> bioinformatics is that a few of the operations (okay, one: sequence
> alignment) are computationally hard.  It's the only reason anyone
> thinks the field isn't trivial.  So we have a big Linux cluster next
> door running a batch submission system called LSF.  You basically give
> it arguments the way you would an exec call, but instead of running it
> locally, it queues it on the cluster, runs it, does some tracking of
> it, and sends you an email when it's done.
>
> I want to take the local executions I'm doing and be able to submit
> them as batch jobs to the cluster instead.  My initial idea was along
> these lines:
>
> 1. Create a temporary directory (the execution will have one anyway,
> just use that).  The scratch space where I would do this is shared
> across the cluster.
> 3. Executions are wrapped in little language in my system, so have the
> macro instead create a uniquely named function to postpone the
> execution.
> 2. Dump an image of the current session into the temporary directory.
> 3. Submit a batch job to start Racket, load the dumped image, and run
> the postponed function.
>
> This is wasteful, but the waste is negligible compared to the size of
> the computing task that would usually be submitted this way.
>
> Unfortunately, I can't figure out how to dump images from Racket.  Is
> it possible?  Or am I having brain creep from Common Lisp?  Or if
> anyone can think of a cleaner way to do this, I would really love to
> hear it.
>
> --
> Frederick Ross
> Bioinformatics and Biostatistics Core Facility
> Life Sciences, EPFL
> http://bbcf.epfl.ch/
> +41 21 693 14 39
> _________________________________________________
>  For list-related administrative tasks:
>  http://lists.racket-lang.org/listinfo/users
>



-- 
Jay McCarthy <jay at cs.byu.edu>
Assistant Professor / Brigham Young University
http://teammccarthy.org/jay

"The glory of God is Intelligence" - D&C 93


Posted on the users mailing list.