Juri Memmert wrote:
> Am Sam, 2002-09-28 um 20.52 schrieb Albertina Lourenci:
> Hi Albertina,
> > I do believe you will appreciate this excellent technical report
> > forwarded (one I also e-mail to the aosd list yesterday from the
excellent research team from PUC-Rio de Janeiro advised by Lucena)!
> Thank you. I read this one a few months back and while I appreciate the
> effort they made, I have some problems with the conclusions... I'll
> write down some of my problems here... maybe they're of use to you.
> > Let me know your impressions!
> Coming right up...
> I'll skip over the preliminary and technical paragraphs for today (I'm
> sick with a cold, so my strength is limited today) and will focus on the
> paragraph 5.
> If you want more feedback on any other chapter or elaborations on my
> comments, let me know.
> "About the base-aspect dichotomy"
> They quote Skipper on this to say that, maybe, possibly, base-aspect
> might be an interesting thing.
> Unfortunately, this is not the case.
> If you have a base-aspect dichotomy, that means that your base can not
> act as an aspect. This limits reuse.
> If you need shared vocabulary, as they call it, that is what declarative
> completeness is about.
> The composition rules of base-aspect, as seen in AspectJ, are limited
> and not very powerful. Bringing base-aspect to MDSOC would simplify the
> composition... and weaken it. Bad idea.
> Explicit dependency introduces high coupling between a base and its
> associated aspects. Bad idea.
The base level based on OO paradigm especially the on class-based language is
too coarse! Imagine when we think of it in terms of generativity! Then it is
For me separation of concerns begins at the base level. And the problem with
doing this through the base level is the same as to insist that all of semiotics
is reduced to the conception of the sign! More modern semioticians such as
Greimas make evident there are levels above the sign and below the sign!
To trace a parallel with OO community and say that above the level lies
the design patterns and the component based paradigm is not ideal, because
the way they are conceived is not generative like a true natural language!
Then to assume that the introduction of separation of concerns would somehow
be a proxy for entities below the level of the sign is equally equivocated!
Since recently Dirk Bäumer and James Noble, Robert Biddle have introduced
Humberto Eco's semiotics into the fore of discussion, I do think Eco's semiotics
may play a role here. In Dirk Bäumer's PHD Thesis Software-Archteckturen für
Rahmenwerkbasierte Konstruktion grosser Anwendungssysteme
he shows the interplay in software of content and expression. According to
the Danish semiotician Hjelmslev a natural language can be understood as
being pervaded by two planes, the content and the expression and he subdivides
each plane according to the strata substance of the content and form of the
content and substance of the expression and form of the expression. I will not
delve deeper into it here because those interested can follow a better
in my next Scientific Report III, ready by October 20, 2002 and available in
my web pages!
You, Juri sees the dynamic viewpoints that can be taken like Dirk Bäumer!
I mean the need to aspectize the base level! Or to see lower level entities
below the class, component, object level, whatever you wish!
In Bäumer's thesis, he even sees in the context of a software architecture
design patterns sometimes as connectors and sometimes as components!
This play with words is avoided when we apply an in-depth analysis through
tools that would avoid this such as the one from Hjelmslev's and the deepest
from Greimas, who is a Hjelmslevian follower! The big difference between the
two is that GReimasian project is GENERATIVE!
Henc for me separation of concerns means to make explicit the network of
relationships that is involved in each single word as Greimas does. It is like
to implode the atom and see it consists of thousands of smaller particles!
Imagine class-based languages are at the level of the molecule!
The implosion started by prototype-based languages like Self really
reproduces better than class-based languages the evolutionary behaviour
of the world! Yet it is still at the sign level!
Then the issue is: How to go from Shannon's notion of passive information
to post-quantum physicist Bohm's active information! Or how to realize
that "God does play fair dice with the quantum universe, but God loads
the dice with consciousness in the post-quantum universe (Jack Sarfatti,
Bohm puts foward:
Passive information may in fact be regarded as a limiting case in which we
from the activity of information. This is essentially the kind of information
that is currently
used in information theory, e.g. as used by Shannon. The puzzle in this approach
of how information that is merely passive within us is able to determine actual
processes outside of us. We suggest that passive information is rather like a
reflecting something of these processes which can guide us to organize them
conveniently for our use, e.f., by means of algoritms that enable us to
But Bohm wants to introduce a concept of active information: We have in this way
introduced a concept that is new in the context of physics -- a concept that we
call "active information"...The basic idea of active information is that a form
very little energy enters into and directs a much greater energy. The activity
of the latter
is in this way given a form similar to that of the smaller energy.
Well this is the basis of generativity! Or how to go from classical information
machines where both the informer and the informee are classical devices to the
mind-brain system (the mind is a quantum device and the brain a classical device
and of course soul is a post-quantum device!) Furthermore the two-way realtion
between the thoughtlike quantum wave and the rocklike classical brain beable
that each of them are simultaneously working as informer and informee. This sets
a self-referential Godel loop which Douglas Hofstadter calls a strange loop and
Erich Harth calls a creative loop. Consciousness requires such a post-quantum
self-determining adaptive loop on the edge between classical determinism and
quantum indeterminism. The capacity of Godel loops to jump out of their own
into strange territory where none have gone before is the nonalgoritymic
of human understanding that ROger Penrose describes in his book Shadows of the
What I want to convey here is that since Gregor is prone to extend AO paradigm
to widen horizons and reach the core of active information, that is in essence
generative, it is not possible to do so without realizing as architects are
their first day of classes at college: God created the universe and the
are the SemiGods that must transform the crust of MOther Earth into building
It is not possible to make a revolution in current computer science without
infusing it with the notion of active information that is a synonym of mimicking
God in exercise of His/Her creative function!
> "About the transformation"
> The AspectJ pointcut model is good. No doubt about that. Removing the
> knowedge imbued in the pointcut definition makes the code less
> expressive. But they did not strive for a high cohesion between the
> classes that embody an aspect. Then, the "aspect-classes" are more
> Furthermore, they resorted to mere Hyper/J. In conjunction with Cosmos,
> much of the lost expressiveness of the pointcut specs is regained in a
> more universal manner.
> Some of the simplifications they did not achieve, they did not achieve
> because they had to reduce badly written aspects to cleanly written
> Java. That caused them some problems. I consider that no problem,
And Java is not a highly expressive and dynamic language at all!
> "About the composition mechanisms"
> They want dynamicism and neither Hyper/J nor AspectJ can provide that.
> The reason why, exactly, they _need_ this seems not convincing. But I
> may be wrong...
If this goes to the AOSD list, I would appreciate it if you could edit
The composition rules of base-aspect, as seen in AspectJ, are currently
limited and less powerful than those in Hyper/J. Introducing base-aspect
to MDSOC would simplify the composition by adding assumptions about base
or aspect. That would, as far as I can tell, reduce the power of the
composition. Bad idea.
Then one is hearing second law of thermodynamics instead
of following fourth ecological law of thermodynamics that informs
you about the trend of the universe to complexify despite the
second law of thermodynamics whose base is essentially generative!
The same to you!
> Name: signature.asc
> signature.asc Type: application/pgp-signature
> Description: Dies ist ein digital signierter Nachrichtenteil