Re: ORBit2 release problems [Re: ORBit2 foo ...]

[michael meeks <michael meeks novell com>]

Hi Mark,

On Tue, 2005-09-06 at 13:06 +0100, Mark McLoughlin wrote:
>   - There has been no release from ORBit2 HEAD

	Right.

>   - Michael seems to suggest that the plan was to ship GNOME 2.12 with 
>     ORBit2-2.12 (i.e from the gnome-2-10 branch)

	Well - of course, I'd have preferred to have been more clued up & done
the releases of HEAD & got that more widely tested, but at this stage -
yes, sounds best.

>   - Because most maintainers have been using jhbuild, and hence jhbuild 
>     HEAD, tarballs released with orbit-idl generated code are using
>     ORBIT2_MAYBE_CONST and won't build with ORBit2-2.12.3

	Right - it's really sub-optimal not having people test what will be
released of course. Having said that - there were some quite substantial
changes in ORBit2 HEAD vs. gnome-2-10 - inasmuch that Tor re-wrote the
basic marshalling code through which ~everything goes - so, most likely
there are some lurking / lingering Solaris/ PPC / odd-endianness
problems there; we'd only see on release.

>   1) Release ORBit2 from HEAD and use that for GNOME 2.12

	Slightly risky at this stage - IMHO. I should do a new development
release though so at least it's not forgotten for the next release.

>   2) Put ORBIT2_MAYBE_CONST on the gnome-2-10 branch and release 
>      ORBit2-2.12.4
>
> 	(2) is probably the best option because ORBit2-2.12 is the better
> tested codebase.

	Right - it's not a controversial or complex change.

	Mark - any chance you can unwind / do a new ORBit2 release ?

	HTH & Thanks,

		Michael.

-- 
 michael meeks novell com  <><, Pseudo Engineer, itinerant idiot