Darren J Moffat wrote:
Shawn Walker wrote:
Darren J Moffat wrote:
I'm trying to understand what we actually gain by having any more
complexity than just repository as the terminology and I'm not sure
I get it yet.
Because a repository alone isn't sufficient to express the identity of
^^^^^^^^^^^^^^^^^^^^
That is a security model concept hence my desire to include that in
the discussion.
of the packages contained within (e.g. there could be a large
difference between libfoo from abc co. and libfoo from xyz co.), nor
is it sufficient to provide a mechanism whereby a user can easily
'override' one package provider's packages with their own.
Again an area where manifest signing comes into play - because one of
the goals of manifest signing is allowing "resigning" exactly for
allowing local overrides.
Could you explain this idea of "resigning" for, and why what's being
proposed here is a) insufficient and b) is relevant at all?
The additional complexity beyond publisher and repository (stream) is
necessary because of the concept of being able to switch between
'development' and 'release' software trains where package versions
alone are not sufficient to express software release types.
The publisher and repository part I'm happy with, I've read the
proposal again since my first post. I'm still unsure if stream needs
to be an exposed concept but I'm leaning in favour of it.
I'm confused. I thought you had concerns over the security implications
of this proposal, but now it sounds like the only concern is whether
stream is an exposed concept or not? If so, that's great. If you still
have concerns about the security aspects, could you state them again
please? Also, at least for me, the more direct and concrete you can make
your explanation, the more likely I am to understand it. (For example,
saying "doing X leaves a gaping security hole because someone can
comprimise it by doing x, y, z" will make an impact. Speaking in vague
terms about certificate life cycles is less likely to make an impression
because I don't have the context to know why I'd care or why it's
relevant. I'm happy to learn, but I need to know why I'm bothering to
learn.)
Thanks,
Brock
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
