Re: Plans for 2.8 - GNOME Managed Language Services?

[Mike Hearn <mike navi cx>]

On Sun, 28 Mar 2004 11:31:20 -0500, muppet wrote:
> The important part would be the metadata.  In order to get it right, 
> you'd be best off forcing the C library projects to eat their own 
> dogfood by generating the C headers and possibly even documentation 
> directly from the API specification metadata.  Yes, i realize that's a 
> pipe dream.  :-(

The reason I suggested reflection (g)interfaces rather than a typelib
binary blob is because then objects being bound from dynamic languages
(rather than to) can just have a single implementation of that interface
which maps to the languages reflection APIs and can avoid the overhead of
generating metadata all at once.

Of course one doesn't rule out the other. You could still have a TLB blob
supporting a generic implementation of the interfaces and then have that
blob page aligned so it doesn't cause too much disk churn (or even on ELF
platforms like Linux an entirely separate ELF section).

As for generating C/C++ headers direct from the type library data, I'm not
sure but iirc on Windows COM components written in C/C++ don't come with
headers at all, it's up to the IDE of whatever language/platform you use
to generate them for you. Documentation comes separately. You get the
binaries and some usage instructions, and maybe source if you're lucky
(but for instance it's common for win32 components to be written in a
language where the compiler costs many hundreds/thousands of dollars so
binaries are normally what you work with).