Hi,

> It's generally a good idea to pick some external VM and built on top
> of it - less work, more power, more compatibility with other
> languages.

When you build on top of some architecture you must consider
obsolescence factors.
From Java up today, all platforms has been discarded
and speed of changes has been increased in the last decade.
The recent history confirm that it will continue happening.

IMHO, a better idea is to do not have a VM;
e.g. make all operations requiered to build aplications
in practice (and not in "abstract") manageable by the system itself
(the system modeled in OO way or not... the same applies).

Changing to another semantics can be good for someone
that think that he has nothing to loose (e.g. someone too young
in the industry); or for someone with very "special" needs;
it si not something good to promote for "all the word"
(as was done repeatedly by M$ and other recent companies
selling that something will happen this year :)

In practice, using well known semantics (e.g. smalltalk semantics)
let companies do not loose investment in education (new
companies and companies that use resources of others
do not have possibility to loose, and has the freedom
to try and waste one to 5 years, upto the next version
of a new "refactored" concept :-)

Returning to the no-VM proposal...
Smalltalk is not a LOO, and it is not a VM design,
it is a system that can be changed at any place
(identity is preserved, Smalltalk is NOT it's contents
at a point in time; it is the flow of contents in a sustainable
system through time and guided by external actions
-e.g. human actions-)

In the past, most architectures supporting smalltalk systems
was designed minimizing the primitive operations
but the VM itself was implemented outside (unmanegeable by)
the system because dynamic generation of low-level behavior
was not possible.
Today, there is more power for dynamic generation
of low-level efficient code; and exists the posibility
to reduce the (Smalltalk)VM to minimal expression,
putting most of the operations relegated to the VM
in the past as responsibility of objects in the system.

e.g. It was frecuent, in smalltalk that lookUp and
evaluate operations was solved in one stage as one VM
operation and objects can´t refine/change the VM behavior.
In recent object models desings[*], new implementations
of object architectures where basic operations (responsibility
of "the VM" in the past) managed by objects in the system
are emerging; we have the posibility to extend/refine/change
any basic behavior WITHOUT loosing robust tools
for systems development...

If we do a good work from any Smalltalk platform
removing the VM as much as possible, it will continue
been a Smalltalk and companies that have invested a lot
in the past, promoting and using smalltalk, will not be
forced to return to languages... (nor move to
"scripting" languages)

I know that the paragraphs put here can be void for
someone young (focused in his/her future and not
in what he/she has already done), but I also know that
not all people/companies are happy about loosing their
investments... again.

w/best regards,
Ale.

p.d.: I have copied to Smalltalking list because we use
to talk about this topics, but in Spanish; and I am interested
in reflecting about how we feel when "new" alternatives
are proposed for all-the-world ignoring peculiarities
of people using marginal development alternatives as Smalltalk.

[*] See objectsAsMethods proposal for Squeak (1999?)
available at current Squeak implementation and Moebius
project ( http://groups.google.com/group/moebius-project-discussion?hl=en )




----- Original Message -----
From: "pako" <pavlo.korzhyk@...>
To: "Strongtalk-general" <strongtalk-general@googlegroups.com>
Sent: Wednesday, September 03, 2008 6:50 AM
Subject: Re: Using V8 for other languages


>
> It's generally a good idea to pick some external VM and built on top
> of it - less work, more power, more compatibility with other
> languages.
> But using VM desgined with one language in mind - there should be
> strong reasons for that.
> How about Parrot VM? Dynamic languages designers is their target
> audience.
> And V8 is (for now) aimed at browser developers only.
> --~--~---------~--~----~------------~-------~--~----~
> You received this message because you are subscribed to the Google Groups
> "Strongtalk-general" group.
> To post to this group, send email to strongtalk-general@googlegroups.com
> To unsubscribe from this group, send email to
> strongtalk-general+unsubscribe@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/strongtalk-general?hl=en
> -~----------~----~----~----~------~----~------~--~---
>
>