dbus-glib - removing the generated bus method wrappers

Havoc Pennington hp at redhat.com
Tue Jul 25 07:31:22 PDT 2006


Matthew Johnson wrote:
> On Tue, 25 Jul 2006, Thiago Macieira wrote:
> 
>> Right, I want them for developer use. So it should be full of annotations
>> that are needed to properly generate code in the bindings.
> 
> what about stuff needed to generate them in the glib bindings? or in the
> java bindings? Who generates these and for which bindings?

This is one of the biggest reasons I was suggesting not putting the 
binding-specific code generation tips in the xml.

If we keep the installed xml to only a "typelib" then bindings simply 
have to be able to work with that, which I think is right - if a binding 
allows extra code-generation tweaking, it should have a feature for 
doing that via a second file (perhaps an xml file that gets "merged"). 
But it should always be possible to use some default settings to 
generate something plausible from just a typelib I'd think.

Anyway, otherwise maintainers will constantly be getting patches to add 
annotations for yet more bindings (or worse) to change the annotations 
to suit one person or another. The maintainer of this file only cares 
what's exported conceptually, not how various bindings use it.

> Can we make a decision on this? I really think there should be the
> ability to include documentation in these, at least single-line
> summaries. I think that a very good API resource is the introspection
> data prettified (for example with my XSLT[1]).

Let's not mix the docs thing in with installing the introspection files 
discussion if possible. If someone wants to independently write up the 
issues, propose a spec patch, etc. on the docs front then cool.

Havoc



More information about the dbus mailing list