Re: The XServiceInfo interface ...
- From: Michael Hoennig "(mi)" <mi sun com>
- To: Michael Meeks <mmeeks gnu org>
- Cc: Michael Hoennig "(mi)" <mi sun com>, gnome-components-list gnome org
- Subject: Re: The XServiceInfo interface ...
- Date: Mon, 23 Oct 2000 12:28:26 GMT
Hi Michael,
> ... I think in general mangling
> random letters at the front of interface names is not a good thing :-)
> Perhaps if all code generating interfaces began with 'X' ( why X ? ) and
> all services were plain, then grokking the idl would be easier. But this
> is not the case; either way it's not a big complaint :-)
why 'X', I don't know anymore. Microsoft started with 'I' and somehow we
chose 'X' to be different. And yes, ALL an ONLY our code generating
interfaces start with this prefix. Or what do you mean? Although UNO has
runtime code generation anyway. But that's a different story.
> So, these names are in a different namespace then, oh well.
For the implementation yes, but it makes documentation generation a lot
easier, of names cannot clash even for services vs. interfaces: We just
create files like "com/sun/star/text/XTextField.html" and
"com/sun/star/text/TextField.html" for example. And all automatically
generated links (from syntactical information) are very straight forward.
If both were "TextField", we had a problem here.
> Tell
> me; is the getImplementationName () method used for anything but
debugging
> ? it seems to me rather strange to allow people to break past the
> query interface abstraction and behave differently with specific objects
> with different implementations; or did I mistake what that method
> does.
True, it is mostly used for debugging and testing. We need to know, which
implementation gets tested to address the right person for the bug report ;-)
> Also, does every object implement this XServiceInfo interface, and
> how does your C++ sugar help people do that ? or is it a relatively rare
> interface ?
Yes, XServiceInfo should be implementation by virtually all objects. It
does not need to be implementated by objects which are only existent in a
very limited context or for example just for offering callback
functionality (being themselves registered as a listener). There is
currently no sugar (I like that term) for the XServiceInfo interface, but
it's easy to type down and the developer has to be aware of the services
which are supported by it's implementation anyway.
Michael
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]