On 4/20/07, Raymond Feng <[EMAIL PROTECTED]> wrote:

Hi,

We have code in core-spi/core today to provide scope management for SCA
components. At spec level, it seems that only the Java spec defines the
scope for java implementations. Do we need to generalize the scope concept
for all component implementation types or should we refactor the code into
implementation-java-runtime to support java components only?

Ant, I was under the impression that the scripting component types may
require the scope management. Can you clarify?


Its a good question. I think it will be common for component implementation
types other than Java to want some concept of scope. The
implementation-script module doesn't support this today though, and there's
a really old JIRA about this, TUSCANY-83.

Seeing as you brought this up I went and had a look at the C++ C+I spec and
that does define scope for C++ components (see section 1.2.3) but instead of
specifying the scope on the .componentType side file it uses an attribute on
the implementation element. That seems good, can easily do the same for
script components.

So I guess I'm suggesting we should keep a generalized scope management
facility.

As to where it should go, I'd been thinking scope (and probably most other
things that the specs define Java annotations for) could go in the
implementation-spi-impl I'd suggested in the SPI reorg thread, but no one
has commented on that thread so I'm not sure others think its an idea worth
pursuing?

  ...ant

[1]
http://mail-archives.apache.org/mod_mbox/ws-tuscany-dev/200704.mbox/[EMAIL 
PROTECTED]

Reply via email to