Re: service stereotype thingies

[Michael Hoennig "(mi)" <mi sun com>]

Hi Maciej,

> > BTW: We have several interfaces which deal with services, I
> > think it would be stupid to change those (of cause they have "service" in
> > their identifier) just because somebody does not like the name. If we
> > merge BONOBO and OpenOffice API, we will have hard times to explain to
> > those developers who have existing code on our API, anyway. I don't see
> > any logical reason to make transition even harder.
> 
> Since Sun asked the GNOME project to rename every method in our IDL, I
> do not have much sympathy. :-)

if it where only renames within OpenOffice, I had some problems in 
communicating it as well, but it would be doable. Unfortunately the 
StarOffice API is already frozen AND published AND used EXTERNALLY. 
Additionally, it is much much bigger than the BONOBO API (having about 
2000 types).

> I think you will have to break interfaces anyway, because in GNOME we
> cannot specify services at the IDL level, since that is not OMG IDL
> compliant.

This conclusion is wrong. If we just renamed the idl files containing 
services to *.svc (or whatever), so that where no IDL files anymore, it 
would not break any interfaces. - And probably the OMG could even go for 
a thing like a service, who knows ...

> How about service-spec (or service_spec, or serviceSpec) as a
> compromise. It's close to what you have, but less likely to be
> confusing as to what it means.

What is the specification of an interface in an idl? Isn't that an 
interface spec? So you write "interface spec" at the top of an interface 
spec? No, you simply write "interface". So, what's the point? I have no 
problem, speaking about these specifications (what they are) as 
"interface spec" and "service spec". But I still don't see why an 
interface should be named "{X}ServiceSpecInfo" instead of 
"{X}ServiceInfo" just for this.

	Michael