Sorry I actually meant profiles and not distributions but was unaware
of subprofiles. Will certainly check it out.
So long as metadata in subprofiles is additive it would be exactly
what I imagined.
Sent from my iPhone
On Jan 30, 2009, at 6:37 PM, Michael DeHaan <mdehaan(a)redhat.com> wrote:
Jeff Schroeder wrote:
> On Fri, Jan 30, 2009 at 2:55 PM, Michael DeHaan
> <mdehaan(a)redhat.com> wrote:
>
>> Jared Hodge wrote:
>>
>>>> I roadmap in Cobbler doesn't always mean everything as we always
>>>> take
>>>> good patches and ideas as they come along, but I think it's still
>>>> important to have a few major features for each release.
>>>> Here's my
>>>> plan. As I see major things going on at the moment for the next
>>>> release (I want to start releasing /smaller/ releases more often,
>>>> rather
>>>> than big releases):
>>>>
>>>> -- performance items. Make CLI talk to cobblerd for speedy
>>>> action in
>>>> large configs. Make all operations callable over XMLRPC for
>>>> better
>>>> integration with other apps. (Possibly add better XMLRPC/web
>>>>
>>>>
>>> search)
>>>
>>>
>>>> -- Windows support (linux-ris work) for mixed environments (I'll
>>>> probably end up helping with this too, since it's a rather big
>>>>
>>>>
>>> feature)
>>>
>>>
>>>> Anyone else see any major themes from the RFE list and things you
>>>>
>>>>
>>> would
>>>
>>>
>>>> like to see concentrated on for this release or upcoming ones?
>>>>
https://fedorahosted.org/cobbler/report/1 ?
>>>>
>
> Chris pointed out an email earlier today that I'm not finding right
> now. Basicly someone had an idea of stacking distros something like:
>
>
> # cobbler list
> Fedora-10-x86_64
> Desktop
>
joeblow1.la.int
>
janeblow1.la.int
> web-farm
>
web1.ny.int
>
web1.la.int
> dns-server
>
ns1.ny.int
>
ns2.fra.int
>
> etc, etc. With the current design of cobbler, it seems fairly trivial
> to do this. Instead of Going Distro --> Profile --> System Record
> things would need to be taught to recurse, but it shouldn't be
> impossible.
>
> This could cut down the number of profiles for some setups
> considerably and would be a generally nice feature to have.
>
>
Profiles are already infinitely nestable (via subprofiles), and has
the same output as the above, so I think this solves the above.
As a distro is defined as a kernel+initrd pair, inheriting distros
themselves does not follow for me at the moment.
If you're proposing something different than the existing
subprofiles implementation, how would it be different?
--Michael