Semantics?!

[Leno Am]

I've decided to forward this to enough mail, as an experiment in open discussion.
And one more thing: you say that you can have object orientation in a functional language. Maybe it's true, but you can also have object orientation in C, and eventually in assembly. What I was trying to say is that object orientation is NOT part of a language's semantics, if it does not encourage you to think that way. A type declaration with a function is close to object orientation, but a stateless object is not the same as a stateful object.

 

---------- Forwarded message ----------
From: Leno Am <len...@gmail.com>
Date: Feb 15, 2007 11:09 PM
Subject: Re: About some issues
To: Eyal Lotem <eyal....@gmail.com>

 

I'm not sure I agree.

Semantics is the way we think about something. Or, it is what we think (if you accept the philosophy that things exist only in our mind).

If you think object-oriented it is very different than thinking procedurally. Procedural and yes, also functional, thinking are not naturally human. We don't think about functions naturally. We think of objects and how you interact with them and their state ("I turned the light on").

This is NOT syntax. Syntax is how you express the thought, not the thought itself (either say "mmjuuju!!" or "i turned the light on", that's syntax).

 

Because google are so rama gvoha, The whole wiki is in SVN and you can actually connect with an svn and pull any version you want, so don't worry about any changes you make.

I'm going to eat some canalonni (yes, we really like italian food) so i'll continue later.

 

On 2/15/07, Eyal Lotem <eyal....@gmail.com > wrote:

I hope there's a history cause I removed your questions in Orthogonal
Persistence.
If you remember them I would be glad to discuss them (preferably on Skype)

Also, I have a lot to say about the MultiSemantics...
In large, I think that a lot of what we call semantics today is
actually syntax too!  Like, the difference between procedural and
object oriented, is nothing but naming and location (location being
the association of a method to the "object" or data). You can actually
have "Object Orientation" in a function language (just put the
functions inside the type declarations).

I think most of the semantics you mentioned in the list are thus moot,
and you are actually left with just 2 possible semantics:  Functional
(i.e: stateless), and Stateful.

I think a language can indeed incorporate these 2 semantics together,
by providing functions and actions. Actions can use functions but not
vice versa.
Actions can be the same as he shows in the 2nd subtext demo.

Anyway, come to talk in Skype when you can.
So much to write, but it takes so many more hours than I thought it would!!!

[Noam]

When viewing the previous message, press "show quoted text" to see
lotex' mail and my first reply.

> On 2/15/07, Eyal Lotem <eyal.lo...@gmail.com> wrote:
>
> > I hope there's a history cause I removed your questions in Orthogonal
> > Persistence.
> > If you remember them I would be glad to discuss them (preferably on Skype)
>
> > Also, I have a lot to say about the MultiSemantics...
> > In large, I think that a lot of what we call semantics today is
> > actually syntax too! Like, the difference between procedural and
> > object oriented, is nothing but naming and location (location being
> > the association of a method to the "object" or data). You can actually
> > have "Object Orientation" in a function language (just put the
> > functions inside the type declarations).
>
> > I think most of the semantics you mentioned in the list are thus moot,
> > and you are actually left with just 2 possible semantics: Functional
> > (i.e: stateless), and Stateful.
>
> > I think a language can indeed incorporate these 2 semantics together,
> > by providing functions and actions. Actions can use functions but not
> > vice versa.
> > Actions can be the same as he shows in the 2nd subtext demo.
>
> > Anyway, come to talk in Skype when you can.
> > So much to write, but it takes so many more hours than I thought it

> > would!!!- Hide quoted text -
>
> - Show quoted text -

[eyal....@gmail.com]

Object-orientation is a way to organize the code, its not really a
semantic.
As you said, you can have object orientation in C, so that just
enforces the point that it is not a semantic.
A language may encourage organizing the code properly (in an object
oriented way) but that STILL is not a semantic.
I believe that in Live Programming, organization is less of an issue,
because the code can be dynamically reorganized by necessity. All the
methods that operate on a type A are associated with that type,
without having to have been declared alongside it (like in C or C++).
The names of those methods also don't have to reflect that
association.

A stateless object (immutable) is not the same as a mutable object,
true. And that is a semantic issue.
That's why I think there should be 2 languages that are incorporated
into one environment. Actions can mutate objects and use functions but
not the other way around.