[Yum] gpg sig checking

[Michael Stenner]

On Fri, Apr 11, 2003 at 01:55:08AM -0400, seth vidal wrote:
> well the idea would be that gpgcheck = 1 would be the default in the
> program defaults so if gpgcheck was unset it would default to on for
> each repository (currently it defaults to off)
> 
> Then if a user turned it off then they'd get a warning message when that
> repository was accessed (processed in the config file more likely)
> 
> keys are easy -  just rpm --import publickey
> 
> if you have an unsigned pkg in a repository where things are expected to
> be signed then an error occurs when you attempt to install that pkg.
> 
> does that make sense?

OK, I think I've got it.  If a user/admin wants to add a repository
with:

1) ALL signed packages, they must

  a) set gpgcheck = 0 for that repository

    or

  b) get the key(s) and add it (them) as described above

2) SOME unsigned packages

  a) set gpgcheck = 0 for that repository


[ I've excluded the obvious "insist that the repo manager sign all
packages" and "don't use the repo" ]

If I understand that correctly, then I think this would be a
reasonable change, provided that the error message when a gpg check
fails (whether due to missing sig or missing key (but probably not BAD
sig)) provides clear pointers to how to "fix" it.  That probably means
documenting it well in the manpage (possibly giving it it's own
section with examples) and then pointing to that from the error with
something like "see the GPG CHECKING section of the yum manpage".

I think Troy's idea is also interesting, but if you do that, you might
want to either break it up into multiple options or use more
descriptive values than 0, 1, 2.  Perhaps "require", "check", and
"ignore".  I don't care about the specific names, it's just that 0, 1,
2 might be confusing.  You can always continue to support 0,1 for
backwards comptibility.

Other than that, I like it :)

					-Michael
-- 
  Michael Stenner                       Office Phone: 919-660-2513
  Duke University, Dept. of Physics       mstenner at phy.duke.edu
  Box 90305, Durham N.C. 27708-0305