[Json] Coming to better consensus on draft-ietf-json-text-sequence

[Paul Hoffman <[e-mail]>]

Greetings again. The WG seems unable to come to consensus on the choice of the record separator in draft-ietf-json-text-sequence. The recent poll finds that there is not even rough consensus about whether using the LF is better or worse than using the RS character in the current draft, and yet people on both sides feel vehement on their choice.

However, the earlier threads about this document show some common ground. There is definite WG interest in log files with sequences of JSON texts, but no agreement in the result as it stands. There was also early statements about the protocol being more complex than was needed, with procedures for receivers dealing with partial texts and truncated entries.

If the protocol is focused on just system log files (not hand-typed files), and without trying to deal with partial texts and truncated entries, the protocol becomes much simpler and the choice of record separator becomes much clearer.

Given this, the chairs ask the document author to produce a new draft:

- Move the first paragraph of Section 3 into the introduction.

- Change the description in Section 2 to be:

    JSON-sequence = ws *(JSON-text ws RS ws)
    RS = %x1E
    ws = <given by RFC7159>
    JSON-text = <given by RFC7159>

- Remove Sections 2.1, 2.1.1, 2.2, and 3. Remove the last paragraph of the Security Considerations.

The chairs believe that such a draft will be more likely to achieve at least rough WG consensus.

--Paul Hoffman and Matt Miller