Re: packaging system

[Mark Wolfe <wolfem@xxxxxxxxxxxxxxxx>]

Sorry about missing the goals section of your site I have since 
revisited it and done some reading. Man there are some pretty good 
ideas there..

Have you had a look at portage in gentoo and how it manages versioning 
as a possible short term alternative to your vision?

On Saturday, July 19, 2003, at 01:11  PM, Matthew Dillon wrote:

> :Gday all,
> :
> :Just a couple of questions.
> :
> :Has anyone got any ideas on what you are envisioning for the packaging
> :system of dragonfly. I have used FreeBSDs ports, debians dpkg and 
> Osx's
> :fink for a bit and I am interested in software distribution and update
> :systems. I would be happy to begin looking into a helping with a 
> higher
> :level design or even just happy to help compile peoples ideas for web
> :content.
>
>     I have a basic idea of what I would like to see, and how it could 
> be
>     accomplished.  I discuss it somewhat in the Goals section of the
>     site.
>
> :Also what are your thoughts of NSS switch.. are you planning to
> :integrate this feature into dragonfly? I am a stalwart supporter of 
> the
> :move to ldap as the core of an os's AAA model.
>
>     Well, I don't know enough about NSS switch to comment on it.  I do
>     know what I want to see for authentication and that is a port 
> service...
>     a user level daemon, which takes and responds to requests from 
> processes
>     for user, group, and other authentication info.  e.g. it would run 
> the
>     password crypt check too, and would be able to ask for (opaque to 
> it)
>     config files and environment variables from the requesting client 
> in
>     order to resolve things like ssh keys, kerberos, and so forth.  It 
> would
>     deal with NIS or other over-the-network authentication systems as 
> well.
>     All of that would be invisible to the requesting client.  I
>     really dislike having to compile authentication support into every 
> program
>     in the system, even if it is in DLL form (like PAM.  I really hate 
> PAM).
>
>     e.g., the conversation would go something like this:
>
>     program: help, I need to authenticate 'charlie'!  I have the 
> following
>     pieces of opaque data:
>
> 	- Something called a ssh2_public_key, whatever that is
> 	- Something called ORIGINATING_IP, whatever that is
>
>     service: send me your ~/.rhosts, ~/.shosts, ~/.ssh/authorized_keys 
> file
>     please.
>
>     program: I only have ~/.shosts and ~/.ssh/... here ya go.
>
>     service: that's good enough, your authenticated for the following 
> (opaque)
>     capabilities: (list of opaque capabilities)
>
>     program: Thanks!  I have no idea what these capabilties are but 
> I'll hand
>     them out (one could be related to ssh that ssh understands.  If 
> this
>     program is ssh then it will understand the ssh-related 
> capabilities).
>
>     And so on and so forth.
>
> :Lastly have you thought about doing some research into some of the
> :technologies used in darwin to possibly add even more to you new
> :distribution. This is just a general fish for ideas from people in 
> this
> :group and is not directed at any particular part of darwin.
>
>     It would depend on the technology.  Some things might not mesh 
> well with
>     the existing goal set, other things might.
>
> 					-Matt
> 					Matthew Dillon
> 					<dillon@xxxxxxxxxxxxx>
>
>
> :BTW good to see there still people out there who are brave enough to
> :break away from the establishment, roll up there sleeves and break 
> some
> :stuff in the name of learning and innovation.
> :
> :Regards,
> :
> :Mark Wolfe
> :Hammond Street Developments