Brainstorm for who to load schema

[Yang ZHONG <le...@gmail.com>]

Tuscany InterOp / JIRA 505 (http://issues.apache.org/jira/browse/TUSCANY-505)
brings us an interesting topic:
there're code or users interested in loading XML, then who is responsible to
load schema?

Here lists some thoughts JUST as a START, it's much more important if YOU
brainstorm please.

I can think of 3 options so far:
3-1. code or users who to load XML, loads schema themselves
3-2. SCA loads component interfaces (WSDL) related schema automatically
3-3. SDO LOCATES and loads schema automatically

3-1 may not be very friendly/efficient since the schema loading burden is
pushed to users who may need to track loaded or not to avoid unnecessary
multiple loading if SDO impl hasn't made such effort.
3-1 may not need to change (SDO) spec.

3-2 might waste effort if component impl isn't SDO based or shcema is not
used on some execution paths.
3-2 may need to change SCA spec.

3-3 may be ambiguous if user deploys conflicting schemas although which may
not be seen often. My previous comercial product automatically LOCATES,
loads and REFRESHES schema on demand.
3-3 may need to change SDO spec.

Really hope to see your comment on any of the options even more options from
you.

Thanks in advance.

-- 

Yang ZHONG

Re: Brainstorm for who to load schema

[Jeremy Boynes <jb...@apache.org>]

On 6/30/06, Yang ZHONG <le...@gmail.com> wrote:
> Tuscany InterOp / JIRA 505 (http://issues.apache.org/jira/browse/TUSCANY-505)
> brings us an interesting topic:
> there're code or users interested in loading XML, then who is responsible to
> load schema?
>
> Here lists some thoughts JUST as a START, it's much more important if YOU
> brainstorm please.

Comments inline ...
>
> I can think of 3 options so far:
> 3-1. code or users who to load XML, loads schema themselves

I think this always need to be available as an option - sometimes it's
the application that knows what to do rather than the infrastructure.
There are APIs for this in the SDO; we provide the @SDOHelper
annotation in SCA that allow them to be injected (to avoid the
.INSTANCE anti-pattern).

I think we could also enhance the introspector to detect properties
whose values are SDO helpers and could eliminate the need for the
annotation. This would be one of the introspections contributed by the
SDO databinding.

> 3-2. SCA loads component interfaces (WSDL) related schema automatically

and not just from WSDL...

I alos mentioned on the TypeHelper thread that we should provide a
mechanism that allowed components to import SDO definitions in their
component-type definition.

> 3-3. SDO LOCATES and loads schema automatically

I don't think SDO should do this. Specifically, I mean the basic SDO
implementation should not do this - it may be a feature of some
framework that is using SDO (like Tuscany't SCA framework).

<snip/>
--
Jeremy

---------------------------------------------------------------------
To unsubscribe, e-mail: tuscany-dev-unsubscribe@ws.apache.org
For additional commands, e-mail: tuscany-dev-help@ws.apache.org