You can view in Wave here:
https://wave.google.com/a/wavesandbox.com/#restored:wave:wavesandbox.com!w%252Bsaq8mHMU%2525A

Or read transcript below:

Office Hours: 2009-08-20

Welcome to Federation office hours!
*Please top post your question or comment about federation to make
sure we see it. You can do so by replying to this blip.*
*A transcript of this wave will be made publicly available. If you
don't want to appear in the transcript, delete your blip after getting
a response. Be careful not to remove neighboring blips.*
*And don't add robots to this wave!*
Aug 20

James Purser:
Sigh, you would do a Federation Office Hours the one day I'm at a
conference wouldn't you ;)
Aug 20

Dmitry Unkovsky:
I'm in hope someone will go here and answer a question ;)
It happened once :)
Aug 20

James Purser:
what's the question?
Aug 21

Dmitry Unkovsky:
It's one blip lower, about extended operations in OT.
Aug 20

Dmitry Unkovsky:
Probably, I'm late here.
Are there any plans to allow extending of OT rules for other types of
data?
Aug 20

Jochen Bekmann:
What kind of data were you thinking of ?
Aug 20

Dmitry Unkovsky:
Sure, current operations defined on xml are quite universal. And quite
any data structure can be represented as xml.
But the question is -- is it always possible to map user interactions
to OT well?
For object like, say, counter, we can't increment it "atomically" with
defined OT operations. (For two users incrementing simultaneously
value of 3, "insert"-"delete" sequence gives us 4 instead of 5 -- just
in idle example)
example is rather trivial, but probably there are more.

I may think of graphs with edges, nodes, and operations "modify-edge"
and "modify-node".
The other example can be moving elements within document.
If I get it right, when one user edits some subelement B of element A,
and the other user moves element A to some other place in document
with sequence of "delete","insert" operations,
editions of element B by first user made after deletion of A get lost.

But sure, for text-based applications currently defined operations
represent user interactions pretty well.
Did you considered possibility of extended operations? It may prove
useful for some applications, was there any discussion on this matter
at google?
Aug 20

Dmitry Unkovsky:
technically, there could be some mechanism for negotiating sets of
operations available to federating servers in manner similar to xmpp
capabilities discovery.
Aug 20

Vadim Barsukov:
Perhaps Dmitry already offline. But the question he had already
discussed a bit in our group - 
http://groups.google.com/group/google-wave-russia/msg/c88947e55ad21c05
(russian). I think he is talking about XML.
Aug 21

Maxim Vasiliev:
handling custom XML is somwehat wider question.
i raised it also in wave-api group:
http://groups.google.ru/group/google-wave-api/browse_frm/thread/f931c8373ea53830
without any sucess, beside consideration:
current API cannot handle xml in any way.
Aug 20

Tom Dyer:
Hi, What plans does Google have about persistance in the Fedone code?
Much talk has been made about a database backend, but I have not seen
much evidence of progress.. Cheers
Aug 20

Jochen Bekmann:
In the short term we do not plan to add a db backend, but have heard
of projects aiming to do so.
Aug 20

Tom Dyer:
OK, Cheers.
Out of interest, do you know if this sandbox implimentation has
persistance, or is it relying on always staying up?
Aug 20

Jochen Bekmann:
The sandbox does have persistance.
Aug 20

Tom Dyer:
Is it in a form that you feel is elegant enough to use on Fedone?
So, is the sandbox running very different code to Fedone?
Aug 20

Jochen Bekmann:
Yes, sandbox is running the full production version of Google Wave, we
have discussed elsewhere (e.g. see in this wave below) that it will
take a little while before the code is open sourced. Our current
database backend will possibly not be easy to open-source as it's in-
house technology for now. It should not be too onerous to port it
though, and code written for FedOne will certainly be appropriate for
use with the production version, provided it uses a production-quality
db.
Aug 20

Vadim Barsukov:
Again, I question. What happens if a wave server - federal
participator -  stop its work? History of all the waves with its
participation will be lost to the participants of this server? Or they
will be able to access them by logging on another server?
Aug 20

Vadim Barsukov:
Sorry, I just want to know - over time or the wrong question?
Aug 20

Jochen Bekmann:
Generally the waves will be lost. There will be copies in wave servers
that are being federated with, so the data will be around, but so far
we have not defined a protocol for how this data might be used to
revive lost waves.
Aug 20

Vadim Barsukov:
If we are still working, I ask another question. How to find in the
not-published waves? For example, on the corporative wave server.Will
the content of the  waves indexed with search engines, for example,
Google? :) Or we need to connect a special robot to every this
conversations?
Can a wave server to work as a local search engine? Search only in
their own waves.
Aug 20

Jochen Bekmann:
Non-published waves will not be searchable as it's private data. You
can search in your own waves. Published waves may be searchable at
some point.
Aug 20

Vadim Barsukov:
Search in the non-published wave is only possible through the client?
But the amount of information can be very large.
Aug 20

Jochen Bekmann:
That is a question for the API team, though I do expect that given the
proper user authentication, it will be possible to search one's waves
via the API.
Aug 20

Vadim Barsukov:
Yes, thanks. I have myself understood ("question about the API").
Thank you!
Aug 20

Casey Whitelaw:
Chiming in: Search is a completely separate piece of functionality.
You can think of a search engine as an agent, with either normal or
privileged access to waves. You can build a wave search engine which
is just a normal web server (not a wave client), etc.
Aug 20

Vadim Barsukov:
Thank you for your clarification. Search in the waves - even a very
interesting challenge, given the more dynamic wave character.

Will there be a special search API? Ohh, OK - this question for the
another day. :)
Aug 20

Vadim Barsukov:
I will ask another question - perhaps naive. :) Can a wave "managing"
the program-client? That is, contain the commands (robots? gadgets?)
that change the status of the browser. For example, add items to the
menu, or open windows, etc.
Aug 20


Sam Thorogood (and Jochen Bekmann):
In the scope of Federation, sure - since a client might allow robots
to do that. However, in the scope of Google's implementation, there
are only a limited set of actions that a robot can perform and AFAIK
they are only on wave content and not on browser state.
Aug 20

Vadim Barsukov:
Limitations in the scope of Google's implementation - this is a
conceptual solution or just for the time being?
 There are actually some parts of the UI that are using content stored
in waves, though it's not quite clear what all you'd include in
"browser status" - window management, for example, is not controlled
using waves, but things like the navigation panel content and other
window content are (or will be).
Aug 20

Vadim Barsukov:
If your browser offers this feature (eg, XUL for Mozilla, or JS-
plugins in Chrome), it can affect the status of your browser and get a
more complete integration of the waves in the client (or even the
operating system, such as Google Chrome OS). Or is it too wild
fantasy? :)
Aug 20

Vadim Barsukov:
Here is a similar idea - 
http://groups.google.com/group/google-wave-api/browse_thread/thread/d1b99258e54fd94c#
Aug 20

Jochen Bekmann:
We like wild fantasy. Though I'd say right now I'm not aware of plans
such as these.
Aug 20

Vadim Barsukov:
Hello! Today we will discuss only the protocol and federation?
Questions about the server FedOne today can be discussed?
Aug 20

Benjamin Kalman:
Yes, questions about FedOne are good.
Aug 20

Vadim Barsukov:
In this case, I have a question. But it is not technical but
organizational. :) Many wondered whether there is a plan for its
development. Will it be brought to a productive state, or will remain
only a demo version? Is there a roadmap?
Aug 20

Jochen Bekmann:
We do have a plan for releasing a production version of our wave
server, however we have not put a date on it - currently we are
working hard on getting our servers solid in order for us to onboard
large numbers of users. As soon as we feel we are done with that work
we will be able to bring them into a state suitable for release. I
apologise for not being able to give you dates.

For those who are keen on running wave servers, we envisage the demo
version should be in a pretty good state soon to be used as a basis
for a production server by people in the community.
Aug 20

Andrew Charlton:
How will the embed API work with federated servers, i.e. will I be
able to embed waves from my own servers in my own intranet
applications without having to connect to Google?
Aug 20


Sam Thorogood (and Jochen Bekmann):
Not initially. We should explain that Google Wave runs on many
different servers, and that the core server, the "wave server" is what
we are federating. The bits that support the browser-based client are
quite separate from that, and is what is needed for embedding. We do
hope to open-source that code too, but it's a while off. There are
open-source projects working on clients which may also allow
embedding, so that might be something of interest while we work on
getting our code ready for open-sourcing.
Aug 21

Maxim Vasiliev:
in wave-protocol group as well as in wave-api there were raised a
couple of questions about exact structure of content document and
operations.
wave protocol draft in no way specifies it.
the last one related to "type" parameter of element and of
startElement operation. FedOne handles only one type="line", FedOne-
client raises exception unless type is "line".
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Wave 
Protocol" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/wave-protocol?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to