[bind10-dev] Minutes, was Proposed agenda for BIND 10 call on 2009-09-23

[Shane Kerr]

All,

Attendees:

- Michael
- Shane
- Jeremy
- Kambe-san
- Jinmei
- Larissa

On Thu, 2009-10-01 at 15:57 +0200, Shane Kerr wrote:
> 
> Agenda for the BIND 10 team call:
> 
>       * Greetings

Jelte away, CNNIC away, Fujiwara-san away, Larissa a bit late today

>       * Agenda bashing
>       * Next week's call 
>               * Shane & Larissa at RIPE, holiday in China

Michael volunteered to take minutes.

>       * Face-to-face meeting details

AP: Shane to arrange hotels

topic: agile methods

>       * How to evolve from prototypes

Shane: thoughts about that?
Jinmei: concentrating on message API

Shane: someone other than prototype finishing it?
Michael: should have something reviewed before we go to production
model, so part of basic task...
Jinmei: point of proposal prove documentation sufficiently easy to
understand... usually have someone review the code to get familiarity
Jinmei: may be beneficial, we are a small group, everyone has their own
tasks, so concerned about efficiency of development with that approach
Michael: me too

Jinmei: can do something like that in the middle of the face-to-face
meeting maybe. Have someone look into someone else's code more
intensively.
Michael: if we discover 6 months later that something was a bad idea we
can still change it. We should put things into production when close to
ready.

Michael: do need to decide on layout of build system, headers go here,
libraries go there, and so on
Shane: is this something in Jeremy's area?
Jeremy: not sure...
Shane: we need something written about this, a proposal at least
Michael: in BIND 9 it was all C, so very easy

Shane: reluctant to go forward without experience in Python wrappers
Michael: assume like we have BIND, single library directory
Jinmei: single library?
Michael: yes, for now
Shane: be "Agile" in the sense that we solve the problem we have today?
Michael: yes
Jinmei: yes

AP: Jeremy will post beginning of this to mailing list

Shane: how to move forward... complete current prototype, or work on
something else first, then come back?
Jinmei: depends on development status
Jinmei: need to understand how this will look like for other components
too. can still discuss and consider ideas for data source API while
working on and completing the DNS message API.
Jinmei: probably send some ideas, but not work on code


Michael: critical we have something working. We have to have something
like a data source. First version is read-only, not SQL, and so on. How
much work do we want to do?
Jinmei: are you talking code for data source?
Michael: would rather see what we need now

Larissa: how does this work for code stability and testing?
Michael: surprised to find out how much guessing we do to figure out
what we need, and then guess wrong

Jinmei: you are proposing we have a simple data source API within this
year so we can have a tiny implementation of a DNS server by end of this
year, for example?
Michael: yes
Jinmei: if we have resources for that I think it makes sense

Jinmei: in case of data source, simple version earlier better


Shane: design by contract in BIND 10?
Jinmei: we should do something similar, but quicker recovery processes
from that

[ long discussion - Michael thinks it's a bit too much, should make
assertion when error discovered ]

>       * Component status/discussion
>               * c-channel (Michael)

Michael: Ruby code done, Python code coming along - learning curve, Ruby
code first did help, hoping to have something checked in today/tomorrow

>               * DNS Message API (Jinmei)

Jinmei: completed initial prototype, send DNS query using new API,
interested in whether rendering part makes sense (in terms of what
Michael pointed out before). Would be good if Michael can look at it.
Michael: will read messages today

>               * Statistics (Kambe-san)

Kambe: no additional message

Jinmei: planning on commit code under experimental directory?
Kambe: yes, we are planning

>               * bob (Shane)

No work.

>         Jelte and CNNIC away, so no configuration or BigTool.

>       * AOB

None.

--
Shane