Re: incomplete languages

Re: incomplete languages

[Karel Miklav]

Dmitry A. Kazakov wrote:
> Alexandre E. Kopilovitch wrote:
>>"Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de> wrote:
>>>A need in a meta language indicates that the core language is incomplete.
>>>Then the question is only whether this incompleteness is inherent or not.
>>>As long as the former is not proved I will claim that we do not need a
>>>meta language.
>>
>>But it is well-known fact (from the first half of 20 century, thanks
>>Godel), that every language that is rich enough is necessarily incomplete.
> 
> Incomplete for what? If a language should be complete for everything, then 
> meta language is also no answer, because it in turn will be also 
> incomplete.

I have this problem: I'm designing an object database, objects are 
directly stored in it. There's no need for SQL as I can load objects 
from disk and check them, update them etc in native language. All is 
good and well, until I move my database on a remote computer. Then, how 
am I supposed to do queries? My database (object store actualy) is dumb, 
downloading a whole database contents for each query is not a solution, 
so I have to export logic to the remote database server. This logic can 
be written in Ada like syntax, but clearly not in Ada, the language.

So; Ada can be compiled, but compiled objects cannot be remotely 
exported. My thoughts went in a direction of meta language or meta 
platform, as it looks like averybody else's. But since David looks so 
shure there must be some other way to export logic remotely (I mean 
without double coding and/or meta languages (while we're here: what is 
Ada, a language or a platform?))?

Regards, Karel Miklav

Re: incomplete languages

[Alexandre E. Kopilovitch]

"Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de> wrote:
> > every language that is rich enough is necessarily incomplete.
>
>Incomplete for what?

Precisely that. I mean that this is the point: if, say, some generic construct
corresponds to a general notion that belongs entirely to the problem area (but
has various specialization or flavors) then there is no reason to climb into
meta-language. But if a generic construct is designed specifically for reuse,
then turning to meta-language may be essential for adequate description, because
"reuse" itself is surely a meta-notion. 

> > If a language should be complete for everything, then
> > meta language is also no answer, because it in turn will be also
> > incomplete.

Certainly yes, meta-language will be also incomplete "for everything".
But with it the base language may appear complete. For example, the base
language may be complete for any particular programming case; assistance
of suitable meta-language makes it complete for more general issues (design
considerations, program comparison etc.); but further generalizations (well,
investigation of methodologies of program comparsion -:) may reveal cases that
can't be adequately described with our meta-language.


Alexander Kopilovitch                      aek@vib.usr.pu.ru
Saint-Petersburg
Russia

Re: incomplete languages

[Dmitry A. Kazakov]

Karel Miklav wrote:

> 
> Dmitry A. Kazakov wrote:
>> Alexandre E. Kopilovitch wrote:
>>>"Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de> wrote:
>>>>A need in a meta language indicates that the core language is
>>>>incomplete. Then the question is only whether this incompleteness is
>>>>inherent or not. As long as the former is not proved I will claim that
>>>>we do not need a meta language.
>>>
>>>But it is well-known fact (from the first half of 20 century, thanks
>>>Godel), that every language that is rich enough is necessarily
>>>incomplete.
>> 
>> Incomplete for what? If a language should be complete for everything,
>> then meta language is also no answer, because it in turn will be also
>> incomplete.
> 
> I have this problem: I'm designing an object database, objects are
> directly stored in it. There's no need for SQL as I can load objects
> from disk and check them, update them etc in native language. All is
> good and well, until I move my database on a remote computer. Then, how
> am I supposed to do queries? My database (object store actualy) is dumb,
> downloading a whole database contents for each query is not a solution,
> so I have to export logic to the remote database server. This logic can
> be written in Ada like syntax, but clearly not in Ada, the language.

Isn't JGNAT and similar things for that? [if JGNAT were continued of course]

> So; Ada can be compiled, but compiled objects cannot be remotely
> exported. My thoughts went in a direction of meta language or meta
> platform, as it looks like averybody else's. But since David looks so
> shure there must be some other way to export logic remotely (I mean
> without double coding and/or meta languages (while we're here: what is
> Ada, a language or a platform?))?

IMO each language is more or less a platform.

-- 
Regards,
Dmitry A. Kazakov
www.dmitry-kazakov.de

Re: incomplete languages

[Georg Bauhaus]

Karel Miklav <karel@inetis.spppambait.com> wrote:
: so I have to export logic to the remote database server. This logic can 
: be written in Ada like syntax, but clearly not in Ada, the language.

Couldn't you, practically, split your program into partitions, one
running on the database computer, using Ada from the Distributed Systems
annex?

-- georg

Re: incomplete languages

[John R. Strohm]

"Georg Bauhaus" <sb463ba@l1-hrz.uni-duisburg.de> wrote in message
news:avvoa2$dbo$1@a1-hrz.uni-duisburg.de...
> Karel Miklav <karel@inetis.spppambait.com> wrote:
> : so I have to export logic to the remote database server. This logic can
> : be written in Ada like syntax, but clearly not in Ada, the language.
>
> Couldn't you, practically, split your program into partitions, one
> running on the database computer, using Ada from the Distributed Systems
> annex?

It sounded to me as though he wanted that built into the language, so that
the programmer would neither know nor care whether the program was focused
in one CPU or distributed across many, by an unspecified communications
subsystem.

Re: incomplete languages

[Karel Miklav]

John R. Strohm wrote:
> "Georg Bauhaus" <sb463ba@l1-hrz.uni-duisburg.de> wrote in message
> news:avvoa2$dbo$1@a1-hrz.uni-duisburg.de...
> 
>>Karel Miklav <karel@inetis.spppambait.com> wrote:
>>: so I have to export logic to the remote database server. This logic can
>>: be written in Ada like syntax, but clearly not in Ada, the language.
>>
>>Couldn't you, practically, split your program into partitions, one
>>running on the database computer, using Ada from the Distributed Systems
>>annex?
> 
> It sounded to me as though he wanted that built into the language, so that
> the programmer would neither know nor care whether the program was focused
> in one CPU or distributed across many, by an unspecified communications
> subsystem.

An application server like web server or an SQL database has to accept 
data and code from its clients. It can parse scripts, but it generally 
has no clue about compiled programs and binary objects it receives, 
unless they are described in another language like IDL, ODL, WSDL or 
something.

If everything could be just written once, that would be perfect; but are 
  compiled languages reflexive enough for something like that? The 
answer is no, by definition; but with some binding technology... Isn't 
it true, that we can express everything in english?

Regards, Karel Miklav

Re: incomplete languages

[tmoran]

> An application server like web server or an SQL database has to accept
> data and code from its clients. It can parse scripts, but it generally
> has no clue about compiled programs and binary objects it receives,
> unless they are described in another language like IDL, ODL, WSDL or
> something.
  A server is a program.  It can be as dumb or as smart as you make it.

Re: incomplete languages

[Karel Miklav]

tmoran@acm.org wrote:
>>An application server like web server or an SQL database has to accept
>>data and code from its clients. It can parse scripts, but it generally
>>has no clue about compiled programs and binary objects it receives,
>>unless they are described in another language like IDL, ODL, WSDL or
>>something.
> 
>   A server is a program.  It can be as dumb or as smart as you make it.

You're right. It could understand your scripts, you could even make him 
to somehow understand compiled Ada code but never a custom serialized 
object - without a binding language. And Ada, the language does not 
provide it, as far as I'm informed. I'm only learning but the closest 
thing I've seen so far is GnatCom. It would be nice if something like 
that was built into Ada.

Regards, Karel Miklav