Hello,
It seems we have had difficulty having any DSpace 2.0 meetings since the 2.0
funded activity has come to an end. I will say that I think we need to find
a way to continue holding meetings. At this point, without continued
activity and buy in from the DSpace community, the 2.0 work risks going the
way of the DAO prototype. Certainly, without a developer active and present
working on the body of code contributed by Aaron / CARET there not a
significant enough activity around the direction of this work to keep it
going and reaching a state of usefulness in DSpace.
At the moment theres was a fairly strong point made by Graham concerning our
ever using dspace-services in a clustered environment. I suspect that if we
cannot clear this imposition, we won't be using the last years worth of 2.0
work for much of anything because everytime it is proposed to be added to
into the tooling for dspace distirubtions, there will be questions about
clustering.
I would like to see a meeting focused on discussing the cacheing and
clustering organized within the community by those stakeholders that use
clustering and designed the 2.0 caching mechanism. I.E. Graham and Aaron, to
resolve this issue so that the general community can utilize this codebase.
Ideally, I'd like to see some clarity on what is distributed across
clustered servlet session, what is to be retained in thread local and why we
need Caching at all in DSpace 2.0 by default. Here are the significant
areas I see needing clarification:
1.) there is a clear difference between retaining the existing "request"
object and cache of objects being used in the request cycle and retention of
any objects outside of that cycle, shouldn't there be separation of request
scope caching, session scope caching and application scope caching to
support our needs, including in the cluster environment.
2.) The current strategy for inclusion of services into dspace 1.6 maintains
the cache within the scope of the webapplication, thus there is one
Kernel/Cache per webapplication rather than one Kernel/Cache for more than
one webapplication. It would be highly beneficial for someone to do an
analysis of the two approaches and how they impact caching and clustering
requirements in dspace-services.
We need to clear these issues and document the architecture or the
application of these tools will be limited to a very distant future.
Finally, in the last developers meeting, I did give concessions to not using
the "dspace-services" in the 1.6 release and am rethinking this conclusion.
I believe, with a little bit of discussion and research around the above
topics, this is an issues that could be resolved and be a win-win for the
community.
Sincerely,
Mark Diggory
------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now. http://p.sf.net/sfu/bobj-july
_______________________________________________
Dspace-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dspace-devel
