SIPCLF - Friday 1pm (Karlin 2/3)
================================

draft-ietf-sipclf-problem-statement-05
Gonzalo Sanguilero presented slides:

Open issue: Make reason-phrase mandatory?
Larger issue: Allow ANY header to be logged?
Issue to be taken to the list.

Chair: Does anyone see a reason why this cannot go to WGLC after final issue resolved?
No opinions expressed.


draft-ietf-sipclf-format-01
Gonzalo Sanguilero presented slides:

Open issue: If no optional fields, TLV start pointer is set to 0x0000.
Would be easier to make it point to the terminating 0x0A.
No dissent expressed - proposed solution to be adopted.

Open issue: Flags are in 'Index Pointers' section, which is otherwise pure metadata.
Move Flag Field to Mandatory Fields (from Index Pointers) to completely separate metadata from 'real' data.
No dissent expressed - proposed solution to be adopted.

Open issue: Sent/received flag is overloaded
Proposed to separate the flag into two - sent/received and transport.
Adam Roach: You know how to use brackets in grep?
Gonzalo: Yes - this issue was on the list.
Robert Sparks: So what?  This is deck-chair rearranging.
Peter Musgrave: Keeps separate things separate.
Robert Sparks: Disagree that grep is a motivating usecase for changes.
Brian?: The current way looks like we are saving bytes, which is not the point of this format.
Meeting agreed to separate into two bytes as proposed.

Open issue: Should we separate encryption from plain text?
Brian?: Not a fan of upper case/lower case distinctions.
Gonzalo: Are there many DTLS signalling implementations?
Adam Roach: This is going the wrong way - don't assume it won't happen
Meeting agreed to separate into two bytes as proposed.

Open issue: current src/dest address representation is not ideal for ipv6
Proposed to log as [address]:port (both IPv4 and IPv6)
Robert Sparks: This will break attempts to reuse existing parsers for IPv4 addresses
Agreed to add [] for IPv6 only

Open issue: Two formats for pre-defined and vendor-specific optional fields
Proposed to use single TLV format.
Agreed to go with list consensus

Open issue: Repeated optional fields - how to represent?
Peter Musgrave (as person): Prefer separate fields with same tag.
Consensus on separate fields.

Open issue: Optional fields logged in ascending tag order?
John Elwell: Extra effort for log creator, so probably better not to specify it.
Taken to list to confirm unspecified order.

Open issue: Final determination of what other fields could be useful to log.
Will discuss on mailing list.

Chair: Hopes to resolve issues on list then go to WGLC?
No dissent expressed.
Chair: Does anyone expect to need another face-to-face meeting?
No opinions expressed.