[cap-talk] Soap box, OO, XML, language nirvana

[John Carlson]

>
I agree that object references and capabilities are similar concepts.
I'm not sure that OO and POLA are necessarily the same though.  Perhaps
"good" OO is POLA.  How do we develop good OO people?  I struggle
with OO myself, and in someways, the web has grown up very non-OO.
I've had a look at Jed's early web programming, and I can say, it was
nowhere close to OO (Sorry Jed).  Does PHP have OO concepts yet?
I think that Waterken is a good solution, if we can figure out how to
make XML and XSLT object-oriented.  If someone thinks these are
object-oriented, please clue me in.  As for server side, can I override
a method in one JSP with one from another JSP?

I think there is something beyond XML and OO that combines the
two technologies.  That's where we're headed.  Maybe Smalltalk
or Ruby or Python already do these things.  Java relies to heavily
on XML/preprocessing, when it should incorporate a structuring  
language itself
(Object arrays aren't type safe).
> How pinned down are we really?  What would it take to get the (a)
> base capability model up to the point that it could be useful to
> users (e.g. launch POLA applications, share permissions with other
> applications and users using POLA via capabilities)?

I think it will require a better better integration between the network
language/protocol and the client & server language.  Is there any  
hope of .NET
or the Vista API doing this?  When we can transparently move
between the network language and and the client & server language,
I think everything will become much easier.  Thus your browser,
server, and network protocols will all support the same language.
Making references externally visible in the language is important,  
that's what
we've learned from URLs.  With Waterken, a language module can  
essentially
declare it's own URL (possibly relative to some configured item),  
instead of relying
on the file system/app server.

Obviously, there should be competitors in this space, so we will
need to develop transformation/gateways to move between
environments.  This will prevent us from being pinned down AGAIN.

If we can't develop language translation for natural languages,
can we at least do it for computer languages?  Do we need a central
language that everyone converts to and from?  Intentional Programming
raises it's head again.  Or should we compile down into virtual machine
code, and decompile back into high level languages?

Do people think that one method per file is a good idea?  Do people
think that files are good ideas?

John