Re: Why is bonobo-config a 2.0 core library?

[Malcolm Tredinnick <malcolm commsecure com au>]

On Sun, Nov 25, 2001 at 06:03:41PM -0800, Maciej Stachowiak wrote:
> Since nothing in the core uses bonobo-config, and since nearly all
> apps are going with straight GConf for the foreseeable future (even
> Evolution is moving to GConf post-1.0 from what I am told), we should
> seriously consider making bonobo-config a non-core component for 2.0.
> 
> That will mean less code to test for the core release, and will give
> bonobo-config a chance to stabilize and mature before we lock it in as
> part of the platform.
> 
> Does anyone else have specific thoughts on the benefits of retaining
> bonobo-config as part of the core release, or the risks/costs of doing
> so?

Sounds reasonable to me (he says, as if his opinion counts for three
beans). I was slightly swayed by Michael's argument last week regarding
language bindings: if a language binding already wraps bonobo, then you
get configuration stuff for free. Currently, they will have to also wrap
gconf.

I don't know enough about the status of the various bindings to know how
much is still to be done in each case. For Python, there are already
some pretty reasonable bonobo bindings and there are also some gconf
wrappers started (on sourceforge, from memory) that need some attention
but are a good start. So there's no real difference for that language.

My only other comment would be to encourage people not to forget
bonobo-conf, although this is really beyond the current scope of the
release. Even if it isn't in the core platform, reminding people that it
exists and is a way of accessing gconf for those already using bonobo in
their applications will probably lower the amount of things they need to
learn (my thinking is that if you have already mastered bonobo and can
read IDL files, then bonobo-conf is easy, whereas gconf is another API.
My thinking _may_ be flawed).

Cheers,
Malcolm

-- 
Change is inevitable except from vending machines.