One other thing I think could be relevant is benchmarking - using
DirectMemory should improve perfomances and this is not demonstrated nor
clearly stated anywhere, but it is the main selling point of off heap
memory. In the past, when I wrote about preliminary results on my blog [1]
I obtained a lot of attention (and hence moved in here in the ASF).

Basically, how to use the "thing" is good but WHY is better. Too bad I have
not had time to work on this stuff, in the last two years, and it is not
likely changing any time soon.

Ciao,
    R

[1]
https://raffaeleguidi.wordpress.com/2011/01/12/directmemory-benchmark-heap-vs-off-heap-vs-orientdb/


Il giorno ven 22 mag 2015 alle ore 13:05 Tommaso Teofili <
[email protected]> ha scritto:

> Hi Christian,
>
> thanks for your comments, you're definitely right!
> I agree with both sides: general documentation and integrations.
>
> One example of that is I had personally worked on an integration with Solr
> [1] but never published the relevant doc to use it.
> I know Apache Jackrabbit Oak uses DM as an optional L2 cache [2,3], but we
> never managed to get this to the users via docs / website.
>
> So in the end we may try reworking website and docs (not sure who else
> could help, but I could spend some cycles on it), and maybe that would
> help.
>
> Regards,
> Tommaso
>
> [1] :
>
> http://svn.apache.org/viewvc/directmemory/trunk/integrations/solr/src/main/java/org/apache/directmemory/solr/SolrOffHeapCache.java?view=markup
> [2] : https://issues.apache.org/jira/browse/OAK-891
> [3] :
>
> http://svn.apache.org/viewvc/jackrabbit/oak/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/plugins/document/cache/NodeDocOffHeapCache.java?view=markup
>
>
> 2015-05-22 12:23 GMT+02:00 Christian Grobmeier <[email protected]>:
>
> >  Hi Tommaso,
> >
> > I just looked at the DM website and I think one of the biggest issues
> with
> > DM is exactly this: the website, and the docs.
> >
> > Lot of pages are empty, some read very short and not informative. That
> > aside, some questions reader might have are not answered. In example
> here:
> > http://directmemory.apache.org/javascript-caching.html
> > I would suspect there is a security issue with that approach.
> >
> > On the other hand, I couldn't get from the website how DM in general
> works.
> >
> > This might lead to the fact the project doesn't get it's deserved
> > momentum. People simple don't know what it is, how to use it, or what the
> > difference to other solutions like EHCache is.
> >
> > Another idea would be to preprare integrations with existing products. In
> > example, one could show the benefits of integrating it with Struts,
> Wicket
> > or Tapestry. Maybe even provide some kind of connector. This may make
> users
> > of these products to DM users. There might be other opportunities.
> >
> > However if there is simply nobody who is willing to help in these
> corners,
> > then unfortunately there is no other way round than the attic. But I
> would
> > think one should pimp the docs first, as they are really not in a good
> > shape.
> >
> > Cheers
> >
> > Christian
> >
> >
> >
> > --
> > Christian Grobmeier
> > [email protected]
> >
> >
> >
> > On Fri, May 22, 2015, at 12:11, Tommaso Teofili wrote:
> >
> > Dear DM fellows,
> >
> > Apache DirectMemory is not having much activities in the latest months
> and
> > we, the PMC, couldn't push forward further developments to move the
> project
> > forward very much, so we are evaluating moving the project to Apache
> Attic
> > [1].
> >  We, the PMC, do not foresee change on our side on our capability of
> > bringing new energy to the project in the short term so, before taking
> > actions for the move to Attic, we would like to give a last call for
> > contributors who want to take the challenge to help us giving Apache
> > DirectMemory some fresh ideas and contributions and empower our
> community.
> >
> >  So if anyone reading here is interested to contribute, please raise your
> > hands!
> >
> >  Thanks and regards,
> >  Tommaso
> >
> > [1] : http://attic.apache.org/
> >
> >
> >
>

Reply via email to