Re: orbit2 build, orbit1 work_pending()

[Michael Meeks <michael ximian com>]

Hi Brian,

On Thu, 2002-02-07 at 23:14, Brian Craft wrote:
> The lack of work_pending() in 0.5.x appears to be a fairly staggering
> limitation.

	It's really not clear what you want to do with work_pending - can you
expand. Sounds like one of those nasty hacks for jaming disparate
mainloops together.

> Do I have to implement another layer of IPC to use orbit with any
> event driven utility, such as gtk? I.e. keep orbit in its own process?  

	No, you shouldn't have to - it should integrate with the glib mainloop
just fine.

> It looks like orbit2 supports work_pending(). Can orbit2 co-exist with 0.5.x?

	You can have both installed on the same system just fine - and develop
against both (given the right versions) - but you can't link to both
(clearly).

> Do clients/servers written against different versions play nicely together?

	Yes - this seems to work with GConf - again with a recent stable ORBit.

>  So if I write a server with orbit2 (and thus get work_pending())
> will all the 0.5.x clients work with it?

	Yes - GConf does this - but can you expand on 'work_pending' ?

> Also, I've been unable to sort out the build for orbit2. I haven't found linc
> outside of gnome cvs. Building from cvs appears to require unspecified other
> stuff from cvs (I don't want to download all of gnome just for this). And then
> there's glib2.

	If you have a RedHat system you can pull snapshots of glib2, linc,
ORBit2 from red-carpet (http://www.ximian.com) - you don't need anything
from Gnome 2.0 but glib2, linc and ORBit2 itself, and these package are
all parallel install clean.

	HTH,

		Michael.

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