You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@avalon.apache.org by Stefano Mazzocchi <st...@apache.org> on 2004/07/18 18:48:12 UTC
My personal view of Avalon
I started Avalon in 1998 with the idea of describing a framework where
reusability of high level components could be easy and socially
scalable, in order to reduce the energy required to write a server.
The need emerged because we were thinking about how to implement James
and we found out that servers had the same needs all around.
The real reusability of code has been a myth since computers were
invented, the first name of the project was "java apache server
framework", but at one point, after the creation of the project but when
basically just Federico, Pier and myself were on it, us three had a
major fight over something that was, indeed, technical, but now looks
like bikesheding (the fact that none of us remembers what it was shows
how technically important it was)
Because we are very good friends, the fight didn't degenerate, but it
became clear to me that the creation of such a framework was a holy
grail itself, so I started to use the term Avalon, to indicate the
island (never found, some say it's Glastonbury) where King Arthur was
burried.
The name was catchy and it remained.
A mythical place, where everybody would like to go to honor the heros of
the past that brought us here, but that nobody is able to find.
The reason for this is found in my essay about ontology harmonization.
http://www.betaversion.org/~stefano/linotype/news/60/
Avalon cannot be found because it doesn't exist. Why? Because, as I
explain in my other essay about categorization
http://www.betaversion.org/~stefano/linotype/news/61/
those refer in particular to the semantic web, but the same exact thing
can be applied to Avalon, since what Avalon tries to achieve is the
creation of a "one has to fit all" explicit semantics.
The most arrogant and pretencious, therefore abrasive, annoying and
irritating activity of all.
What we all failed to see was that there cannot be a single unitarian
view of the world. Thoughts and categories are not metaphysical entities
that we can discover, but are just tools and technologies.
But unlike some people here tend to think, there is nothing more
objective about technology, or math for that matter (see Roger Penrose,
Shadows of the Mind), than say, emotions or politics.
The more abstract a concept becomes, the less objective discussion can
be, no matter if these abstractions represent technologies.
Avalon is creating a conceptual reference model. In the library
community, this is well known to be the hardest jobs of all. Normally
every effort takes more than a decade and burns out 80% of the people
and, at the end, the one with most social/political pressure imposes its
views.
Believe me: had I known that, there would be no Avalon today.
- o -
Now, is it so bad?
Well, Stephen is trying to be reasonable and is backing off and I
appreciate that, but if you like your bike red and I like my bike black,
we cannot have the same bike.
We can go around all day saying how wonderful it would be if all agreed
on one thing and if there was peace of earth and all that, but reality
indicates that some of the divergencies exist because of the intrinsic
mental differences of people.
Those who refute the computability of the human mind (aka AI), including
Alan Turing himself, indicating in the "ability to disagree and make
mistakes" the ultimate indication of human intelligence.
Now, Stephen, look around: we failed. There are different camps and they
will never agree.
Why? because agreement requires trust and while some people trust you
(or are willing to put up with you because they need your unpaid work)
some other don't and after you burned bridges, and it's hard to rebuild
them, especially if they see no value in doing so.
I see no value in a unified "one size fits all framework". Why? *VERY
SIMPLE*: the concept of reusability of code was done to save time and
energy, nut the process of ontological harmonization requires *more*
effort than simply writing the code.
Result: it's a waste of time.
Cocoon depends on the avalon framework 4.1.5, but we are stopping right
there and we'll follow our own path from now on because that *saves us*
energy.
The idea of saving energy by the creation of a reference model has been
demostrated to be flawed, and not because of technical issues, but
because of human ones.
This is actually a great discovery, IMO.
Avalon has failed.
We must signal that to people very clearly and move on, each one of us
on their own paths.
--
Stefano.
Re: My personal view of Avalon
Posted by Leo Simons <ls...@jicarilla.org>.
Stefano Mazzocchi wrote:
> And IMO, Avalon is not [efficient and socio-economically viable].
Rather harsh statement, don't you think, considering people make money
from doing avalon-based development? You consider them an anomaly?
I think you would do everyone a pleasure if you could differentiate a
little between the "original vision" for avalon and what it actually
(regardless of "new vision"s). IOW replace all the current and future
tense in these messages with past tense and it sounds a lot less harsh.
Regardless, what conclusions do you draw from your "personal view of
Avalon"? Just abandon it all? That seems a bit incompatible with your
suggestion to incubate merlin earlier on. And highly inefficient from a
socio-economic viewpoint ;)
I'd finally like to point out that "efficient and socio-economically
viable" is basicaly a non-goal of the ASF. Gump, for example, is not
particularly good at doing "efficient and socio-economically viable"
things, but is a thriving project nonetheless :-D
cheers,
- LSD, not much of a visionary himself
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@avalon.apache.org
For additional commands, e-mail: dev-help@avalon.apache.org
Re: My personal view of Avalon
Posted by Leo Sutic <le...@gmail.com>.
On Mon, 19 Jul 2004 15:43:54 -0400, Stefano Mazzocchi
<st...@apache.org> wrote:
> At the end, it's a matter of priorities.
True.
So, what would your priorities be for Avalon? Where to from here?
/LS
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@avalon.apache.org
For additional commands, e-mail: dev-help@avalon.apache.org
Re: My personal view of Avalon
Posted by Stefano Mazzocchi <st...@apache.org>.
Leo Sutic wrote:
> On Mon, 19 Jul 2004 12:12:39 -0400, Stefano Mazzocchi
> <st...@apache.org> wrote:
>
>>avalon has showed (at the very least to *me*) that reusability by
>>extreme abstraction doesn't work because the cost of harmonizing human
>>perceptual differences at such high levels of abstraction is higher than
>>the cost of creating the code.
>
>
> My lesson learned is that reusability that comes at the cost of having
> to re-engineer your deployed code doesn't gather much of a following.
> One could say that any project that aims to bring reusability to
> development had better make it extremely easy to reuse whatever code
> already exists. The cost of harmonizing our code bases is higher than
> to just plod along - especially since the target we would be
> harmonizing with is still moving.
>
> In short, it's not the human perceptual differences that keeps us from
> achieving nirvana, it is that we have code and systems to support, and
> no time to dick around with evolving a framework and constantly
> rewriting our existing code.
>
> Not so much social issues as economic.
I think that even if we had all day to rewrite our code, the problems
would still be the same.
In fact, ontology writers just do that all day and give a damn about
back compatibility, since they just aim at perfection and think that you
should be able to sacrifice everything for that.
At the end, it's a matter of priorities.
--
Stefano.
Re: My personal view of Avalon
Posted by Leo Sutic <le...@gmail.com>.
On Mon, 19 Jul 2004 12:12:39 -0400, Stefano Mazzocchi
<st...@apache.org> wrote:
> avalon has showed (at the very least to *me*) that reusability by
> extreme abstraction doesn't work because the cost of harmonizing human
> perceptual differences at such high levels of abstraction is higher than
> the cost of creating the code.
My lesson learned is that reusability that comes at the cost of having
to re-engineer your deployed code doesn't gather much of a following.
One could say that any project that aims to bring reusability to
development had better make it extremely easy to reuse whatever code
already exists. The cost of harmonizing our code bases is higher than
to just plod along - especially since the target we would be
harmonizing with is still moving.
In short, it's not the human perceptual differences that keeps us from
achieving nirvana, it is that we have code and systems to support, and
no time to dick around with evolving a framework and constantly
rewriting our existing code.
Not so much social issues as economic.
/LS
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@avalon.apache.org
For additional commands, e-mail: dev-help@avalon.apache.org
Re: My personal view of Avalon
Posted by Stefano Mazzocchi <st...@apache.org>.
Stephen McConnell wrote:
> Stefano Mazzocchi wrote:
>
>> Avalon has failed.
[snip]
> Visions don't fail - they just change our understanding of what's possible.
Fair enough, allow me to rephrase:
avalon has showed (at the very least to *me*) that reusability by
extreme abstraction doesn't work because the cost of harmonizing human
perceptual differences at such high levels of abstraction is higher than
the cost of creating the code.
Very true, this is indeed not a failure but a great lesson, but it
didn't change my understanding of what's possible, it changed my
understanding of what's efficient and socio-economically viable.
And IMO, Avalon is not.
--
Stefano.
Re: My personal view of Avalon
Posted by Stephen McConnell <mc...@apache.org>.
Stefano Mazzocchi wrote:
> Avalon has failed.
Prototypes, trials, experiments (successful and unsuccessful) in the
United States together with research acquired from Germany paved the way
to Chuck Yeager's mach 1.1 flight during October 1947. Some people said
it was impossible. Others embraced the notions of 'further', 'faster',
'higher' and through conviction, arrogance, bravery, collaboration, and
the death of thirty-six test pilots - their goals become a reality.
Within twenty-two year man was walking on the moon. A little over a
year later, Gene Kranz utter the words "failure is not an option" - and
thirty-four years later we are sending probes to Mars.
Visions don't fail - they just change our understanding of what's possible.
:-)
Cheers, Steve.
--
|---------------------------------------|
| Magic by Merlin |
| Production by Avalon |
| |
| http://avalon.apache.org |
|---------------------------------------|
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@avalon.apache.org
For additional commands, e-mail: dev-help@avalon.apache.org
Re: My personal view of Avalon
Posted by Stephen McConnell <mc...@apache.org>.
Leo Sutic wrote:
> On Sun, 18 Jul 2004 12:48:12 -0400, Stefano Mazzocchi
> <st...@apache.org> wrote:
>
>>if you like your bike red and I like my bike black, we
>>cannot have the same bike.
>
>
> True.
>
> The bane of reusability is usually that you have two different
> scenarios that only differ in a minor detail - but that detail happens
> to be spread throughout the source code.
>
> But remember this:
>
> + We all use the same language (Java)
> + We run with the same standard libraries
>
> So reusability *isn't* a myth. Its just that you can't reuse
> everything. So Stefano, how much can be reused? What, in your opinion,
> can be salvaged from Avalon? Do we just flush everything down the
> drain? Or keep some things?
>
>
>>at the end, the one with most social/political
>>pressure imposes its views.
>
>
> That is my experience too.
>
>
>>Avalon has failed.
>
>
> Depends on definition of failure. In the objective "provide a unified
> server framework", yes. But there are some useful code in Avalon (if I
> could keep just one set of classes, I'd keep the o.a.a.f.configuration
> package), and I think it has helped introduce IoC and all that for
> many people (myself being one of them).
>
> So we didn't reach architecture nirvana, but we got somewhere.
>
> So when you say "move on, each one of us on their own path", it can't
> mean "dump every piece of Avalon code in your system and just move
> on", but something else.
>
> What?
Hi Leo:
If you subtract Fortress and Merlin from the Avalon picture your left
with framework, logkit, the cornerstone suite. If this is the Avalon
that we are talking about then your talking about some good stable
pieces of technology used by lots of people in lots of projects. I
don't see it evolving but I do see the need for its maintenance and long
term housekeeping.
Cheers, Steve.
--
|---------------------------------------|
| Magic by Merlin |
| Production by Avalon |
| |
| http://avalon.apache.org |
|---------------------------------------|
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@avalon.apache.org
For additional commands, e-mail: dev-help@avalon.apache.org
Re: My personal view of Avalon
Posted by Leo Sutic <le...@gmail.com>.
On Sun, 18 Jul 2004 12:48:12 -0400, Stefano Mazzocchi
<st...@apache.org> wrote:
> if you like your bike red and I like my bike black, we
> cannot have the same bike.
True.
The bane of reusability is usually that you have two different
scenarios that only differ in a minor detail - but that detail happens
to be spread throughout the source code.
But remember this:
+ We all use the same language (Java)
+ We run with the same standard libraries
So reusability *isn't* a myth. Its just that you can't reuse
everything. So Stefano, how much can be reused? What, in your opinion,
can be salvaged from Avalon? Do we just flush everything down the
drain? Or keep some things?
> at the end, the one with most social/political
> pressure imposes its views.
That is my experience too.
> Avalon has failed.
Depends on definition of failure. In the objective "provide a unified
server framework", yes. But there are some useful code in Avalon (if I
could keep just one set of classes, I'd keep the o.a.a.f.configuration
package), and I think it has helped introduce IoC and all that for
many people (myself being one of them).
So we didn't reach architecture nirvana, but we got somewhere.
So when you say "move on, each one of us on their own path", it can't
mean "dump every piece of Avalon code in your system and just move
on", but something else.
What?
/LS
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@avalon.apache.org
For additional commands, e-mail: dev-help@avalon.apache.org
Re: My personal view of Avalon
Posted by hammett <ha...@uol.com.br>.
Howdy!
----- Original Message -----
From: "Stefano Mazzocchi" <st...@apache.org>
> those refer in particular to the semantic web, but the same exact thing
> can be applied to Avalon, since what Avalon tries to achieve is the
> creation of a "one has to fit all" explicit semantics.
Stefano, what has failed is the single unified view as a solution to all
problems in the world, not Avalon itself. You post sounded too harsh, and
might sound even more terrible to people who has investments on it - like
myself.
As you probably know not all PMC members at the time believed in the unified
vision, but after some flame fest we decided to step out and leave the way
open for those who pursued this dream. I can't see why Avalon team couldn't
review this position now.
Cheers,
hammett
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@avalon.apache.org
For additional commands, e-mail: dev-help@avalon.apache.org