Eclipse Community Forums: EMF "Technology" (Ecore Tools, EMFatic, etc) » [CDO][0.8.0] Accessing more directly the data

Help

Home

Home » Modeling » EMF "Technology" (Ecore Tools, EMFatic, etc) » [CDO][0.8.0] Accessing more directly the data

Show: Today's Messages :: Show Polls :: Message Navigator

[CDO][0.8.0] Accessing more directly the data [message #96617]

Wed, 12 September 2007 22:43

Simon Mc Duff

Messages: 596
Registered: July 2009

Senior Member

In our processing our services runs on the same machine than the CDOServer.

We would like to access the data more directly.. and we really do not need
to go through the transport layer.

Also in this specific case, we do not need to have a cache in
RevisionManager and another one in the CDORevisionImpl.

Could we replace CDORevisionImpl by the RevisionManager ?

I know right now we cannot... but if all the operation are hidden behind an
interface... We should be able to it.

Request for a view will need to be hidden behind and interface.

P.S. This should be easy to do ... once we can have mutliple server for the
same repository.

Report message to a moderator

Re: [CDO][0.8.0] Accessing more directly the data [message #96647 is a reply to message #96617]

Thu, 13 September 2007 09:08

Eclipse User

Originally posted by: stepper.sympedia.de

Simon McDuff schrieb:
> In our processing our services runs on the same machine than the CDOServer.
>
> We would like to access the data more directly.. and we really do not need
> to go through the transport layer.
>
There is another solution in Net4j to bypass the socket laye:
You could use JVMConnectors instead of the TCPConnectors you are
currently using.
JVMConnectors don't bypass the complete transport layer but they pass
IBuffers from
one IChannel peer to the other through direct method calls. This way the
threading model
of the receiver is the same as with any other IConnector implementation
but a bit of your
performance is lost (compared to no transport layer at all, which is not
the business of Net4j)
due to buffer multiplexing.

Note: Currently the JVM connector implementation could be broken since
I'm only using
the TCP connectors for my current tests.

> Also in this specific case, we do not need to have a cache in
> RevisionManager and another one in the CDORevisionImpl.
>
If you think we should add a configuration option to bypass caching
please file a feature request.

> Could we replace CDORevisionImpl by the RevisionManager ?
>
I guess you mean something different from what you actually typed ;-)
You certainly don't mean CDORevisionImpl here. Maybe CDORevisionManagerImpl?

> I know right now we cannot... but if all the operation are hidden behind an
> interface... We should be able to it.
>
> Request for a view will need to be hidden behind and interface.
>
> P.S. This should be easy to do ... once we can have mutliple server for the
> same repository.
>
I'm not that sure that it's very simple.
Chances are much higher that I will be able to look at this soon if you
attach patch with a proposal to your respective feature request.

Cheers
/Eike

Report message to a moderator

Re: [CDO][0.8.0] Accessing more directly the data [message #609706 is a reply to message #96617]

Thu, 13 September 2007 09:08

Eike Stepper

Messages: 6693
Registered: July 2009

Senior Member

Simon McDuff schrieb:
> In our processing our services runs on the same machine than the CDOServer.
>
> We would like to access the data more directly.. and we really do not need
> to go through the transport layer.
>
There is another solution in Net4j to bypass the socket laye:
You could use JVMConnectors instead of the TCPConnectors you are
currently using.
JVMConnectors don't bypass the complete transport layer but they pass
IBuffers from
one IChannel peer to the other through direct method calls. This way the
threading model
of the receiver is the same as with any other IConnector implementation
but a bit of your
performance is lost (compared to no transport layer at all, which is not
the business of Net4j)
due to buffer multiplexing.

Note: Currently the JVM connector implementation could be broken since
I'm only using
the TCP connectors for my current tests.

> Also in this specific case, we do not need to have a cache in
> RevisionManager and another one in the CDORevisionImpl.
>
If you think we should add a configuration option to bypass caching
please file a feature request.

> Could we replace CDORevisionImpl by the RevisionManager ?
>
I guess you mean something different from what you actually typed ;-)
You certainly don't mean CDORevisionImpl here. Maybe CDORevisionManagerImpl?

> I know right now we cannot... but if all the operation are hidden behind an
> interface... We should be able to it.
>
> Request for a view will need to be hidden behind and interface.
>
> P.S. This should be easy to do ... once we can have mutliple server for the
> same repository.
>
I'm not that sure that it's very simple.
Chances are much higher that I will be able to look at this soon if you
attach patch with a proposal to your respective feature request.

Cheers
/Eike

Cheers
/Eike

----
http://www.esc-net.de
http://thegordian.blogspot.com
http://twitter.com/eikestepper

Report message to a moderator

Previous Topic:	[EMF Compare]Comparing own model (Dynamic Emf)
Next Topic:	[EMF Compare]Comparing own model (Dynamic Emf)

Goto Forum:

-=] Back to Top [=-

[ Syndicate this forum (XML) ] [

]

Current Time: Thu Jan 02 17:14:59 GMT 2025

.:: Contact :: Home ::.

Breadcrumbs

Sign up to our Newsletter