incubator-directmemory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Raffaele P. Guidi" <>
Subject Re: DirectMemory in OSGi.
Date Wed, 19 Oct 2011 22:22:04 GMT
Amazing stuff going on here :-) Decision has been taken to keep the existing
singleton facade but adding new APIs beside it when needed - like in this
case. Just keep in mind that memorystorage is also a singleton and a new
facade has to be provided for it as well. Just go ahead as you feel and
let's see how it comes up - code talks ;-)

On Wednesday, October 19, 2011, Ioannis Canellos <> wrote:
> Hi,
> I couldn't resist on assigning
> my self.
> I added everything that was needed to make directmemory-cache an OSGi
> bundle. All of our runtime dependencies (direct & transitive) are
> as OSGi bundles so its quite easy to run directmemory inside OSGi.
> I also added a new module for integration tests of directmemory in OSGi so
> that we can test its runtime behavior. Currently there are pretty basic
> stuff in there but more will be added soon.
> I am a bit concerned however, about the Cache object and how it
> itself. The fact that it acts as a singleton (not in the strict way) can
> cause issues we running inside OSGi. From some integration tests, I run I
> think that the Cache cannot be shared between different bundles, so each
> bundle will have its own Cache. We need to deal with it if we want to
> have direct memory running inside OSGi.
> I would suggest that instead of having all these statics, to have
> like a CacheFactory that could be used to create a Cache object, then we
> could export this object as a service and have a shared Cache for all
> bundles to use?
> What do you think?
> PS: I am really new to direct-memory, so please forgive me if I miss
> something obvious here.
> --
> *Ioannis Canellos*
> *
> FuseSource <>
> **
> Blog:
> **
> Apache Karaf <> Committer & PMC
> Apache ServiceMix <>  Committer
> Apache Gora <> Committer
> *

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message