[DAIS-WG] DAIS RDF Telcon Notes

[Malcolm Atkinson]

Rob et al.

real distributed systems with autonomously evolving components are the  
only ones we will ever have on any large and long-term scale.
No one can freeze services, data types, representations, standards,  
etc. for the life-time of systems.
Ergo, that complexity is fundamental to large and long running systems.

It is also fundamental to would-be component and service providers  
that they will have to operate (a) in a great many different contexts,  
and (b) in contexts that change.
Ergo, whether we like it or not, we have to face that complexity.

Most DB research, most Software Engineering research, most DB research  
is virtually useless because it pretends we can magically avoid  
autonomous change by some unspecified trick or suggestion completely  
incompatible with human and organisational behaviour.

One of the reasons DBs have acted as the glue-ware for coupling  
distributed application systems is because the *commercial* products  
have some support for autonomous evolution: applications have the "I'm  
the only kid on the block" illusion via transactions, schema editors,  
queries that deliver results of the same type even when source tables  
have changed, views, distributed queries, etc. Not as elegant,  
efficient or flexible as we would like but a good pragmatic solution.

I'm totally happy for exploration of alternative underpinning  
mechanisms for OGSA-DAI because it makes it more resilient to  
contextual change and introduces more paths to adoption.  But every  
distributed-systems engineer must realise that the complexity  
generated by autonomous is real and develop strategies for coping with  
it. RESTful hasn't as far as I know, so it will all be down to the  
individual developers and system architects.  WS has, in principle,  
that in mind, but stalled for reasons described in another email.

Malcolm


On 16 Jul 2009, at 12:20, Rob Baxter wrote:

> Hi Isao,
>
> It's something the OGSA-DAI team has started to explore in their  
> "spare
> time".  Personally I think anything that can alleviate the tooling
> incompatibilities/config pain/encoding overheads of WS-* has to be a
> good thing.  I've hated webservices for a long time now...
>
> Cheers,
>     Rob
>
> Isao KOJIMA wrote:
>> Hi,
>>
>> For RDF, I agree with Oscar.
>>
>> However, I think this is same issue with all WS-DAIs.
>>
>> In this sense, I personally think HTTP GET/POST-based
>> restful subset(or profile or bindings or something like that)
>> of WS-DAI/R/X might be useful, especially for Relational.
>>
>> Is anyone of the list interested in this issue?
>>
>> Regards,
>>
>> Isao
>>
>> -----------------------------------------------------
>> Isao Kojima
>> Leader, Service-ware Research Group
>> Information Technology Research Institute
>> AIST,Japan
>> --
>>  dais-wg mailing list
>>  dais-wg at ogf.org
>>  http://www.ogf.org/mailman/listinfo/dais-wg
>>
>>
>
> --
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Dr Rob Baxter - Software Development Group Manager
> EPCC - University of Edinburgh
> phone: +44 (0)131 651 3579  mobile: +44 (0)7971 437 749
>
> "When I became a man I put away childish things, such as the
> fear of childishness and the desire to be very grown up" - C.S. Lewis
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>
>
> -- 
> The University of Edinburgh is a charitable body, registered in
> Scotland, with registration number SC005336.
>
> --
>  dais-wg mailing list
>  dais-wg at ogf.org
>  http://www.ogf.org/mailman/listinfo/dais-wg
>