From: Ilias Lazaridis <ilias@...>
Date: 2005-02-20T20:39:43+09:00
Subject: Re: [EVALUATION] - E02 - Nitro, a Ruby Based WebFramework

Austin Ziegler wrote:
> On Sun, 20 Feb 2005 07:14:43 +0900, Martin DeMello
> <martindemello@yahoo.com> wrote:
> 
>>Ilias Lazaridis <ilias@lazaridis.com > wrote:
>>
>>>>Supported (FastCGI, Postgresql, Oracle)
>>>
>>>This answer referes to RDBMS. I have corrected the document to
>>>contain both (RDBMS, OODBMS).
>>>
>>>http://lazaridis.com/case/stack/index.html#deployment 
>>>
>>>I assume at this point that nitro does not support OODBMS
>>>bindings.
>>
>>Quite apart from Austin's rant on the fundamental brokenness of
>>the OODBMS concept, might there not be an impedance mismatch
>>between the DBMS's notion of an object and the language's?
>>Singleton methods are the simplest example I can think of in the
>>Ruby case, but I'm sure there are others.
> 
> Generally speaking, yes. The only OODBMS systems that will work well
> with a particular language and said language's object model are
> those written in that language. Generally, one won't be able to
> effectivly use a Java OODBMS from a C++ program -- you need to have
> the C++ implementation of the OODBMS.

correction: "you need to hav the C++ binding to the OODBMS".

> The Prevalence model (Madaleine, etc.) is an OODBMS for Ruby.
> There's also a few others, e.g., GOOD.

Note that this is again off-topic.

The (sub-)topic here is: does Nitro (which is _not_ Ruby) support 
large-scale OODBMS bindings?

The answer so far is: no.

-

example OODBMS for Large Scale Distributed Systems, which has bindings 
for C++, Smalltalk, Java:

http://www.objectivity.com/Products/Products.shtml

> -austin

..

-- 
http://lazaridis.com