Re: GEP / C++ bindings ...

[Michael Meeks <michael ximian com>]

Hi Murray,

On Thu, 2002-08-15 at 16:52, Murray Cumming wrote:
> a) Work has been done on a combined ORBit C/C++ project at the expense
> of the separate project.

	I'm curious as to what has been done at the expense of a separate
project - my feeling is that it wouldn't be _that_ hard to split out
some or all of the C++ stuff.

> b) Changing the plan now would cause a lot of delay.

	Possibly, but I can put some time in to obviate that.

> c) Changing the plan now might make a separate ORBit/C++ impossible
> because
>    1. It would need ORBit2 itself to be restructured in a way that is
> not allowed after a freeze.

	Hmm; not convinced - we can add API; nothing the C++ binding adds,
cannot be supported by adding API, it's just whether this is a good idea
- which is not clear.

	Ultimately we get to add API [ the C++ API ], but perhaps if we add a
small C API for you guys, we can keep that more stable - and keep a
large new unstable C++ API/ABI out of ORBit2.

>   2. It would require ORBit/C++ to reimplement lots of
> currently-undocumented low-level code. That would be a wasteful
> duplication of code and effort. And I know it's beyond my abilities.

	Well; how much is there; in fact I think Gergo has done rather a good
job of keeping this code out of sight and down in volume - isn't that so
Gergo ? from a quick read there were not many places. Then again, there
is quite a chunk of stuff exposed in the CORBA_any bits - but then again
we tend to use that C ABI in other places, so it's not really changable
anyway so ...

> So the problem is that merging seems to be the only way to have an
> ORBit2 C++ mapping. If we don't merge then I expect the ORBit2 C++
> mapping to die.

	Well; the thing we have to do is to persuade the relevant maintainers
in public discussion of the gep; I'll post it shortly.

	Regards,

		Michael.

-- 
 mmeeks@gnu.org  <><, Pseudo Engineer, itinerant idiot