And less unfriendly to users, which is an IETF property
", "time": "2022-07-28T14:13:56Z"}, {"author": "Jon Geater", "text": "I very much like the new threat actor wording
", "time": "2022-07-28T14:14:15Z"}, {"author": "Brendan Moran", "text": "I'll try to leave some code for how to do this with COSE in the SUIT repo
", "time": "2022-07-28T14:28:18Z"}, {"author": "Penglin Yang", "text": "thanks very much
", "time": "2022-07-28T14:29:31Z"}, {"author": "Kohei Isobe", "text": "Now TEEP protocol provides some schemes for conveying attestation result/evidenece, but we'd better to make a hint which to use for readers in drafts.
", "time": "2022-07-28T14:59:55Z"}, {"author": "Ken Takayama", "text": "@Brendan Moran
\nsuit-directive-unlink would be there in suit-uninstall
nothing is shared
", "time": "2022-07-28T15:36:30Z"}, {"author": "Jon Geater", "text": "slides showing here
", "time": "2022-07-28T15:37:02Z"}, {"author": "Sorin Faibish", "text": "yes
", "time": "2022-07-28T15:37:04Z"}, {"author": "Brendan Moran", "text": "Ken Takayama said:
\n\n\nBrendan Moran
\n
\nsuit-directive-unlink would be there in suit-uninstall
That's what I expected. Would it be anything at all other than a list of components to unlink?
", "time": "2022-07-28T15:42:17Z"}, {"author": "Dave Thaler", "text": "@Brendan Moran probably not, as long as the install doesn't do anything besides something that unlinking the last reference won't undo
", "time": "2022-07-28T15:45:33Z"}, {"author": "Brendan Moran", "text": "I would expect 4 commands:
\nset-component-index = [a, b, c ,d]\nunlink\nset-dependency-index = true\nprocess-dependency\nagree
", "time": "2022-07-28T15:45:55Z"}, {"author": "Brendan Moran", "text": "Is it possible that there would ever be a required ordering between dependency unlinks and component unlinks?
", "time": "2022-07-28T15:46:33Z"}, {"author": "Brendan Moran", "text": "and/or, conditional unlinks?
", "time": "2022-07-28T15:46:52Z"}, {"author": "Ken Takayama", "text": "@Brendan Moran
\nThat sequence is what I thought.
my opinion: no... because if refcounted, it would only go to 0 when it's ok. i.e.internal refcounts cover any ordering issue.
", "time": "2022-07-28T15:47:17Z"}, {"author": "Brendan Moran", "text": "If there's any ordering requirement or conditional behaviour, then we might need a command sequence
", "time": "2022-07-28T15:47:38Z"}, {"author": "Brendan Moran", "text": "I have mixed feelings. On the one-hand, it strikes me that we only need a list of components to unlink.
\nOn the other hand, if there might be conditional unlinking or there might be an ordering requirement for unlinking between components specified in the current manifest and those specified in a dependency, then we need a command sequence.
\nSimilarly, the spirit of SUIT is to not imply too much, which pushes towards an uninstall command sequence.
", "time": "2022-07-28T15:51:59Z"}, {"author": "Dave Thaler", "text": "I'm not worried about the security aspects any more than I was before, btw. anything that it can do in an uninstall command sequence it could have done in install directives with a new manifest sequence number, so no change in security considerations that I can think of. Any security issue was already there in the install directives.
", "time": "2022-07-28T15:57:31Z"}, {"author": "Ken Takayama", "text": "Brendan Moran said:
\n\n\nIs it possible that there would ever be a required ordering between dependency unlinks and component unlinks?
\n
It should be reverse sequence of suit-install. In case dependee first depending second in suit-install, it should be depending first dependee second in suit-uninstall.
", "time": "2022-07-28T15:57:33Z"}, {"author": "Dave Thaler", "text": "(I voted don't wait,do after)
", "time": "2022-07-28T16:00:01Z"}, {"author": "Sorin Faibish", "text": "We can discuss off line the container usecase
", "time": "2022-07-28T16:00:50Z"}, {"author": "Penglin Yang", "text": "sure
", "time": "2022-07-28T16:03:05Z"}]