"Autocoding Reloaded" (was Re: Matrix - Metaphors, Analogies and the REAL

Peter Hansen peter at engcorp.com
Mon Nov 10 22:53:31 CET 2003


3seas wrote:
> 
> the project is in python code
> http://freshmeat.net/projects/victor1/

That much is true.  I just downloaded and perused the latest set
of code from that site and attempted to determine once again what
it is intended to do and why Timothy has such trouble explaining
it (because his inability to get the idea across fascinates me:
either the idea has no merit but is consuming an ungodly amount of
someone's time, or both the idea and Tim are brilliant yet he 
is amazingly incapable of explaining it in terms we mortals can 
understand.  Either way it's interesting.  No, I'm not interested
in the potential middle ground here: it's uninteresting. ;-) )

This time, unlike the last time I checked a year or two ago, there
_is_ actually Python code present.  Quite a bit of it, Rexx no more...

Interpreting the Python code, then, and building on Tim's past attempts
at explaining, I believe that the basic idea is to implement a 
standard way of performing nine abstract operations which Tim feels 
are fundamental to all computing.  Or all something... maybe life 
itself.  (I'm onclear on the extent of his belief in this area.)

What it actually appears to me has happened, however, is that he
has partially implemented a system which, when complete, would 
operate at roughly the level of A-A-P or Scons, but would do so in
a more roundabout and less usable manner.  (Using the term "usable"
in the technical sense of ergonomics and usability.)  I can't see 
any evidence of anything practical that has been developed using the
existing stuff, nor any hard examples (with sequences of actual 
commands) which would demonstrate any potential beyond this.

I can't see anything there which could not be done more easily and 
succinctly, for the most part, with a few simple lines of Python.
Python is, after all, a general-purpose programming language, so surely
it can be used to solve the task in the first place.  The VIC code
appears on the path to provided a bunch of library routines to support
certain operations, but even at this point it appears a number of them
just wrap existing library routines, and the others are fairly simplistic
things such as I would expect to find in A-A-P or Scons or any other 
make/build tool like that.

Maybe that's not the point: maybe the point is that the standardization
will make such tasks accessible to non-programmers.  If that's the
case, I suppose who can say?  Are there past efforts similar to this
which have not succeeded?  It's hard to argue when the project has still
not delivered anything workable, or good use cases to support the goals.

In my opinion, but certainly speaking as I can only from the point of
view of an experienced programmer, learning to use Python itself would
be far more productive in any time frame than learning the (unfinished)
VIC commands and then attempting to warp each problem into a form which
could then be attacked with the "nine commands which are at the heart of
all things" (not a quote from Tim, just my own phrasing).

I think the fact that Tim has successfully learned Python enough to 
implement what he has implemented, yet still has not got anything practical
out of the VIC concept itself (as far as I can tell) is quite telling.

Clearly Python is the more practical, easy to learn, and valuable of the two.
And it's easy to understand what it is.

-Peter




More information about the Python-list mailing list