Skip to main content

Minutes interim-2020-mls-19: Tue 12:00
minutes-interim-2020-mls-19-202008251200-00

Meeting Minutes Messaging Layer Security (mls) WG
Date and time 2020-08-25 16:00
Title Minutes interim-2020-mls-19: Tue 12:00
State Active
Other versions plain text
Last updated 2020-09-14

minutes-interim-2020-mls-19-202008251200-00
# Issues/PRs

# Issues/PRs

- PR #397
  - **Good to merge.**

- PR #393
  - Richard: Is intent that application messages MUST NOT be plaintext?
  - Raphael: Yes.
  - Richard: I was considering a use-case where we don't use MLS framing,
    we just use MLS to export a key. MLS-SRTP is an example.
  - Raphael: If you don't want confidentiality, you can explicitly put
    your data in the AAD. Right now it's fragile because messages might
    not be encrypted and you might not notice.
  - **Good to merge.**

- PR #396
  - Brendan: This makes it seem that it's ok to not rotate your signing
    key if you can't secure it, when it's not ok.
  - Konrad: Brendan's right that rolling the signing  key solves other
    problems. But we leave a lot of authentication out of the protocol,
    so I can see the value of having the membership token.
  - Richard: I like this because it adds enforcement of the sender.type
    field.
  - Britta: The security properties we focus on currently are also all
    framed in terms of the current group.
  - Konrad: I would feel better if we had an explicit MAC'ing primitive.
  - Britta: There's an authentication secret in another PR, that's
    specifically for this problem.
  - Richard: Would be my preference to derive something new off the key
    schedule.
  - **Changes requested:** use real MAC, and with new new key.

- PR #389
  - Richard: This PR contrasts with #369 about extensions in Commits. Need
    cleaner story about extending commits. I'm ok with proposals being the
    only extensibility method for Commits.
  - Rahael: Sounds viable. Right now you can omit the `path` of a Commit
    if it's only Adds. But if we introduce a PSK proposal, are we allowed
    to omit path? Not clear.
  - Brendan: I think extensions can arbitrarily change the processing
    logic for whatever is appropriate for them.
  - Sean: Can we merge this or do we need more changes?
  - Raphael: I think we need to be clearer on the logic that requires a
    `path` or not.
  - **Changes requested:** Change pseudocode to look for Updates or
    Removes, instead of only Adds.