2008/6/24 Venky <[EMAIL PROTECTED]>:
> On Tue, Jun 24, 2008 at 10:37:10PM -0500, Shawn Walker wrote:
>> 2008/6/24 Moinak Ghosh <[EMAIL PROTECTED]>:
>> > On Wed, Jun 25, 2008 at 8:35 AM, Shawn Walker <[EMAIL PROTECTED]> wrote:
>> >> That's mainly what I'm getting at; just put a much better way.
>> >>
>> >> It's perfectly fine, in my view, to require that the materials
>> >> necessary to rebuild the package be provided (even if those materials
>> >> are primarly binaries in some cases).
>> >>
>> >> There's no reason to require duplicated efforts to have packages in a
>> >> repository.
>> >
>> >   Ok. A suggestion.
>> >   For the longer run I think it is pertinent to evolve 
>> > tooling/infrastructure
>> >   for automated builds of package submissions possibly using the Test
>> >   Farm resources. Essentially the submission of a package triggers the
>> >   scheduling of a future build of the source recipe to verify correctness 
>> > with
>> >   a mechanism to get the results back to the contributor. Human 
>> > intervention
>> >   should not be necessary.
>>
>> Yes, an automated build facility would be great. I just wouldn't want
>> to see us setup a contrib repository where the only way to get things
>> into it was through the intervention of a build team member.
>
> The SFE repository is working very well, so I'd assume that does not
> need to be fixed.  Which leaves the just one major missing piece --
> the pkgbuild-pkg one.

It is working well, yes. But one solution does "not fit all."

>> Users
>> should just be able to contribute a package and necessary materials
>> and have it "show up" (perhaps after a quick manual approval by
>> someone).
>
> Almost the same as what I'm saying.  The only extra step is where an
> automated build is triggered after the approval.  That is the only
> real overhead, and has the major advantage that the approval has
> a much higher probability of assuring quality and security than just
> inspecting binaries.
>
> The alternative, IMO, is the slightly more heavyweight trusted
> maintainer model.

I believe a mixed model is more appropriate.

In short, I'm just going to have to disagree with you on requiring
things to be buildable to be contributable.

-- 
Shawn Walker
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to