Re: service stereotype thingies



[I'm not hacking with bonobo, so please excuse me if I miss something
obvious]

I have some problems about using oaf for checking what services a given
bonobo object supports. I'm not an oaf guru, but I think it is quite
common in object systems to have objects that aren't activated through
oaf. Many objects just get passed to the client and may never been
activated through oaf. I don't know how it would be possible to check
such an object for an service.


btw i don't like the term service, too. We should think about sth.
better (stereotype?).

   martin

Maciej Stachowiak wrote:
> 
> Elliot Lee <sopwith redhat com> writes:
> 
> > It seems as if all that is really needed is to use an OAF query of e.g.
> >       "repo_ids.has_all(['IDL:Intf1:1.0','IDL:Intf2:1.0','IDL:Intf3:1.0'])"
> > This is sort of a pain to use because one has to type out all the
> > interfaces that a "service" is supposed to offer, instead of just using a
> > single name. I am thinking it is doable to add aliases in OAF to turn
> > service names into a list of interface names.
> 
> I was going to suggest the same thing, for two reasons.
> 
> 1) We can't specify services in the IDL because we're using OMG IDL
>    and can't just arbitrarily extend it.
> 
> 2) An OAF query already specifies the same info that a service does
>    (including support for various attributes and optional interfaces.
> 
> > Another way to solve this is just to make sure that each "service"
> > (component, object impl, whatever) has a primary interface associated with
> > it, and if an object implements that primary interface it also implements
> > all the other interfaces associated with that "service". It seems like an
> > interface for each service will be needed anyways in many cases. This
> > approach solves the XCell hack, and requires the least # of changes
> > (none!).
> 
> Yes, looking at it, a service seems rougly equivalent to a contract to
> implement a given set of interfaces, plus a symbolic marker.
> 
> Rough sketch of a plan:
> 
> * OAF could take care of registering service definitions.
> 
> * An oafinfo file for a server could specify what services it
>   implements, in addition to what interfaces; if services are
>   specified as OAF queries (or something that can be turned into one
>   easly), then OAF could even check that a server meets a service
>   contract when reading the .oafinfo file.
> 
> * We should not do any X prefixing, since services will live in a
>   namespace totally orthogonal to interfaces.
> 
> BTW, I am not sure I like the name "service" for this. I also don't
> like the MS COM term, "type". Does anyone have a better name?
> 
>  - Maciej
> 
> _______________________________________________
> gnome-components-list mailing list
> gnome-components-list gnome org
> http://mail.gnome.org/mailman/listinfo/gnome-components-list






[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]