Pre-InterOp VTP draft

[John Swinbank]

Dear TDIG,

In advance of next week’s InterOp[1], I have returned to my efforts to update the VOEvent Transport Protocol note.

As a reminder, the current version (1.1) of the note by Alasdair & Bob resides at <http://www.ivoa.net/documents/Notes/VOEventTransport/>. I have previously sent draft versions of an updated version to this mailing list, most recently in May of last year. My twin goals are to address the issues which I encountered when developing an implementation of version 1.1 and, ultimately, to shepherd the document towards adoption as an IVOA recommendation. 

In addition to minor changes to the wording and formatting of the document, I have made the following more significant revisions:

- Add Section 3.1, describing design goals of the protocol.
- Add Section 8, detailing requirements for message de-duplication to avoid network loops.
- Specify an explicit interval requirement to connection maintenance messages (§5).
- Clarify the semantics of nak transport messages (§6.3).
- Make it explicit that brokers should not attempt to repeat delivery of messages which meet with a nak on the first attempt: VTP does not support the concept of a “temporary failure” (§§6.3, 7.3).
- Reword the descriptions of protocol operation so that they describe only the traffic exchanged over the network and not the implementation of the various entities (§7).
- Allow timezone specification in iamalive <TimeStamp /> elements (§§6.1, 6.2, 6.3).
- Remove identifying information from example XML documents (§§6.1, 6.2, 6.3, 9.2.2).

I have also converted the document from MS Word to LaTeX.

My work-in-progress is available from <https://github.com/jdswinbank/vtp>. From <https://github.com/jdswinbank/vtp/releases> you can download PDFs of both the latest version and a LaTeXed version of the draft sent to this mailing list in May 2013.

Comments and contributions are welcome, either on this mailing list or via GitHub pull request. Perhaps it’s worth pointing out that the aim here is to formalize a description of the protocol which is already in use in the wild in a few different implementations (see http://wiki.ivoa.net/twiki/bin/view/IVOA/IvoaVOEvent#VOEvent_Transport) and which aims to serve as the lowest-common-denominator: I’ve no objection to (indeed, I have some enthusiasm for) discussing bigger, better, shinier, faster, more generic, more functional, etc means of transporting VOEvents, but that’s out of scope for this document.

Cheers,

John

[1] Submissions & requests relating to the TDIG session still very much welcomed!