[bind10-dev] Minutes, was Proposed agenda for BIND 10 call on 2009-09-10

[Shane Kerr]

All,

I had an editor failure in the middle of the meeting (I hit "escape" and
"yes" to the "cancel changes?" dialog), so I lost the minutes from the
beginning of the meeting. Fortunately most of the substantive
conversation was preserved.

Anyway, corrections, additions, or questions can go to the list or to me
directly.


Attendees:

Jeremy
Feng
Jin
Likun
Evan
Shane
Jelte
Kambe-san
Michael
Larissa

On Thu, 2009-09-10 at 15:56 +0200, Shane Kerr wrote: 
> Agenda for the BIND 10 team call:
> 
>       * Agenda bashing
>       * AP from previous meetings
>               * Shane to set up bind10-changes list

Was waiting on Ops to set up the list. [ Update: this has been done ]

https://lists.isc.org/mailman/listinfo/bind10-changes

>               * Shane to summarize code standards discussion and put on
>                 Trac site

Not yet done.

>               * Jin to send recommendation about patterns to list

Not yet done.

>               * Shane will send notes from previous BigTool discussions
>                 to CNNIC developers

Done.

>               * Shane to have a look at statistics overall design as
>                 documented by JPRS and confirm it is correct

Done.

>               * Shane & Larissa will talk to Francis about working on
>                 Python interface to DNS Message API

Not yet done. [ Update: Shane has since talked to Francis about this ]

>       * BIND 10 development meeting (2009-10-26 to 2009-10-30)

Just a reminder.

>       * NETCONF (Jelte and Michael)

[ Note: real minutes begin here! ]

Jelte explains YANG (IETF NETCONF draft).

Michael: Is this too complicated? Why not use SQLite for everything?
Shane: Interesting, we still need notify...
Michael: Yes we have to add some stuff.

Michael: Maybe still use YANG and use it to build classes?
Shane: Could ignore the rest of NETCONF and use YANG as a data modeling
language.

Shane: Is learning curve bad for NETCONF?
Michael: It is there.

Michael: Should use stuff from OpenReg for this stuff.
Shane: Okay, do it.

*** DECISION: We will use OpenReg for the C-channel ***

Michael: event structure of some sort?
Michael: don't want FD here, lock here, and so on
Shane: use file descriptor everywhere (Unix-like - Windows is different)

Shane: So... still look at YANG?
Michael: I think we should look into it
Jelte: Yes
Michael: Why not use some other thing also?

Michael: should use NETCONF externally, maybe use this for BigTool
Jelte: part of what I was originally thinking, if we have to do it in
the end, why not start out with it?
Shane: how hard is client side?
Michael: client side is easy, structure server data to make
NETCONF-friendly

Shane: is NETCONF a hurdle for people writing clients?
Michael: yes, specific clients

Shane: we need to take to the list...
Michael: HTTP means SOAP, REST, XML-RPC

Michael: if we're going to use SQLite for config stuff initially, we
don't need custom formats (like for zone timers)

> * AOB

Shane: Ad-hoc meetings are good!
Jelte: we could do with more discussion
Michael: I'm always available!

Jeremy: point out that test server is not installed yet, a bit
disappointing (been waiting for 4 weeks)

AP: Shane to bring that up at a management level (to Leo) [ UPDATE: the
server is now up ]

Michael: Meetings where sit and talk about design ideas. Also sit in the
same room when coding. Maybe we should be in the same room and coding.
Hackathons.
Jelte: code-sprints... you *really* get a lot done!

--
Shane