Home » Eclipse Projects » Technology Project and PMC » stellation and distributed/discontinued operation
|
| Re: stellation and distributed/discontinued operation [message #9927 is a reply to message #9904] |
Thu, 18 July 2002 12:49  |
 Mark C. Chu-Carroll
Messages: 64
Registered: July 2009 |
Member |
|
|
On Wed, 17 Jul 2002 20:52:55 -0400, Florin Iucha wrote:
> Is support for distributed and/or discontinued operation planned for
> Stellation? What level of work might be required? Are there any small
> steps that would bring it closer to that?
Distributed: the current system supports it. There are two ways to access
a Stellation repository: through a local database, and through a
server. We *strongly* recommend using the server. If you're running
a server, multiple programmers can access it concurrently from all
over the net.
Discontinued: we've called this disconnected operation. We don't do
it yes, but we plan to. For how we'll do it, there are two answers,
because we want to cover two different approaches.
First, we are going to provide an externalization mechanism that
can be used to integrate Stellation with other systems. We'll be
implementing a mechanism that will allow a Stellation repository to be
exported into a simple external form that will allow copies of a
repository to reconcile with one another. Using this, you can put a
Stellation repository inside of a ClearCase view, or use an rsync-like
protocol to reconcile copies, or put it into a WebDAV server and
use WebDAV merge operations to reconcile copies. With this, you can
easily manage a small number of replicas scattered around the network,
which will stay loosely consistent with one another. (By loosely
consistent, I mean the same kind of consistency you get with ClearCase
Multisite - the servers will constantly be reconciling, but there's
a time window in which one server may know about a change that other
servers don't know yet.)
Second, if you look at the papers on our website describing our plans,
we plan to implement a heirarchical repository replication mechanism. With
replication, you can have a repository leaf replica on your personal
machine, which you can use whether or not your connected to
the network. Then, when you are connected to the network, you reconcile
with your parent repository (using exactly the mechanism described
in the previous paragraph). So you are, in effect, always connected to
a repository, but it works whether or not you're connected to the
network.
-Mark
--
Mark Craig Chu-Carroll, IBM T.J. Watson Research Center
*** The Stellation project: Advanced SCM for Collaboration
*** http://www.eclipse.org/stellation
*** Work Email: mcc@watson.ibm.com ------- Personal Email: markcc@bestweb.net
|
|
|
| Re: stellation and distributed/discontinued operation [message #568990 is a reply to message #9904] |
Thu, 18 July 2002 12:49 |
Mark C. Chu-Carroll
Messages: 64
Registered: July 2009 |
Member |
|
|
On Wed, 17 Jul 2002 20:52:55 -0400, Florin Iucha wrote:
> Is support for distributed and/or discontinued operation planned for
> Stellation? What level of work might be required? Are there any small
> steps that would bring it closer to that?
Distributed: the current system supports it. There are two ways to access
a Stellation repository: through a local database, and through a
server. We *strongly* recommend using the server. If you're running
a server, multiple programmers can access it concurrently from all
over the net.
Discontinued: we've called this disconnected operation. We don't do
it yes, but we plan to. For how we'll do it, there are two answers,
because we want to cover two different approaches.
First, we are going to provide an externalization mechanism that
can be used to integrate Stellation with other systems. We'll be
implementing a mechanism that will allow a Stellation repository to be
exported into a simple external form that will allow copies of a
repository to reconcile with one another. Using this, you can put a
Stellation repository inside of a ClearCase view, or use an rsync-like
protocol to reconcile copies, or put it into a WebDAV server and
use WebDAV merge operations to reconcile copies. With this, you can
easily manage a small number of replicas scattered around the network,
which will stay loosely consistent with one another. (By loosely
consistent, I mean the same kind of consistency you get with ClearCase
Multisite - the servers will constantly be reconciling, but there's
a time window in which one server may know about a change that other
servers don't know yet.)
Second, if you look at the papers on our website describing our plans,
we plan to implement a heirarchical repository replication mechanism. With
replication, you can have a repository leaf replica on your personal
machine, which you can use whether or not your connected to
the network. Then, when you are connected to the network, you reconcile
with your parent repository (using exactly the mechanism described
in the previous paragraph). So you are, in effect, always connected to
a repository, but it works whether or not you're connected to the
network.
-Mark
--
Mark Craig Chu-Carroll, IBM T.J. Watson Research Center
*** The Stellation project: Advanced SCM for Collaboration
*** http://www.eclipse.org/stellation
*** Work Email: mcc@watson.ibm.com ------- Personal Email: markcc@bestweb.net
|
|
|
Goto Forum:
Current Time: Sat Nov 09 03:47:18 GMT 2024
Powered by FUDForum. Page generated in 0.03440 seconds |
] 
Search
Help
Register
Login
Home
