[gin-info] Possible list of subset data for info interop

[Thomas Soddemann]

Hi Erwin,

are you aware that you are proposing to use a n^2 solution and 
diregarding a linear scaling one with n being the number of schemas used?

If we just try find that list of attributes and map all schema 
representation onto each other we have a n^2 problem. Here every pair of 
Grid projects will have to map their schemas on top of each other. -> n(n-1)
If we use a minimum schema (whatever that will be) as the interface 
schema representation for exchanging information we only deal with a 
linear scaling problem, since each Grid project will have to map its own 
schema to the interface one. -> n

If we only had two schemas, e.g. Glue and CIM, the task was trivial, but 
if I remember right, we have more than two.

Cheers,
Thomas

Erwin Laure wrote:

> Absolutely! It is always tempting in these efforts to start defining 
> new things because obviously existing solutions have their 
> deficiencies. But that's not the point for GIN. As Jen said, we should 
> try to come up with a minimal set and try to build translators between 
> different representations (Glue/CIM more specifically). That's all GIN 
> should be doing.
>
> Hopefully, this work will lead to some insights that we can feed into 
> standardization processes, or specifically for the information system, 
> into glue2, but we are not defining it.
>
> Cheers,
>
> -- Erwin
>
>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: Thomas.Soddemann.vcf
Type: text/x-vcard
Size: 255 bytes
Desc: not available
Url : http://www.ogf.org/pipermail/gin-info/attachments/20060302/70d95801/attachment.vcf