Why is Dan Bron so sexy ;-)
-----Original Message-----
From: Dan Bron <[EMAIL PROTECTED]>
To: Programming forum <[email protected]>
Sent: Wed, 30 May 2007 10:39 am
Subject: Re: [Jprogramming] converting APL-PLUS sf files
> In particular, could one make the
> workspace-interchange-file format work
> with J?
J's eschews binary formats by design. So, unless you need ongoing
communication
between J and APL+Win, the best course would be, IMO, to calcuate your data
from
primary sources. Whereever and however you acquired the information for the
original APL components, recreate that logic in J.
If that's not possible, then I would convert your data once and store it in
files which J can read natively.
That leaves you two options: ASCII script files, and binary jfiles. Now,
unless your data is very large and takes a long time to "calculate" (i.e. build
using verbs on literal and numeric constants 0 : 0 ), I would use straight
ASCII script files.
Script files are easier to manage: read, understand, change, share, and
version. Their primary drawback is that they're hard to write. That is,
given
an arbitrary J noun, it is difficult to produce an ASCII representation that
has
the above advantages (in particular, the "easy to understand" advantage).
To my mind, the 5!:5 representation of a complex J structure has no
advantages
over the 3!:1 representation. And it's not clear which is more future-proof
(both the definitions of primitives and the internal representation of J nouns
can and do change from one major version to the next).
But, since you said your data has 3 > rank and 0 = L. the 5!:5
representation should be just fine, maybe with a little help from 0 : 0 .
Having said all that: the workspace-interchange-file format is just that: a
format. J is a Turing-complete programming language. Any format readable and
writable by another TCPL (in this case, APL+Win) is readable and writable by J.
All that's wanting is a couple of verbs.
Perhaps, if you point us at the specification for the
workspace-interchange-file
format, we can help you write those verbs.
-Dan
PS: Another interesting data serialization format is YAML: http://yaml.org .
It's standardized, unambiguous, and is a good complement to J. Whereas J is a
wonderful notation for describing operations upon data, YAML is good for
describing the data itself.
It's one of my goals to write a decent YAML library in J. (I will break the
spec a bit by extending YAML's prosaic constant notation to a subset of J's
more
expressive variety)
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
________________________________________________________________________
AOL now offers free email to everyone. Find out more about what's free from
AOL at AOL.com.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
