Hi Vladimir,
I am using Indrajit's scala library, not in ESME, but in another project
project that is a precursor to features I want to later add to ESME.
I am not able to say with any certainty that Lift's Record API doesn't
fit well; I started to think about it by sketching two implementations,
one for documents (ESME nodes, pure get/put key-value pairs where the
value is a JSON representation of the node) and one for timelines (the
Redis List structure), then decided, perhaps if for no other reason than
gross simplicity, to go with simply using the Redis API directly. I'll
offer this hunch: Record extensions can, and perhaps should be
implemented for ESME nodes and for timelines. For now, I'm not going
that direction but don't rule it out in the future.
I, too, am looking forward to seeing the results ;-)
Jack
On 9/13/2011 9:28 AM, Vladimir Ivanov wrote:
Hi Jack!
Your thoughts regarding Redis integration seem very interesting.
Unfortunately I haven't had time to look closer into it. As I understood
from your comment - Lift's Record API doesn't fit well with Redis? Are you
going to use indrajitr's scala library for Redis?
I'm looking forward to see the results!
Vladimir
2011/9/6 Jack Park<[email protected]>
At the moment, I'm not sure about creating a Jira for it, but I am working
on it. I started with
https://github.com/indrajitr/**scala-redis<https://github.com/indrajitr/scala-redis>
and, thanks to conversations with Indrajit, have learned that I can craft
what I call RedisModel (an object) that holds a live instance of RedisClient
and presents a useful API for doing timelines and nodes. I'm working on that
now. RedisClient must be configured in Boot.scala
When I get something that works, I'll toss it up to github and ask others
to polish the work of this scala beginner.
My present thinking is that user modeling is still handled in an RDBMS
(though, I suppose, even there it's possible to run into scaling problems)
Jack
On 9/2/2011 10:25 AM, Ethan Jewett wrote:
That's a good point. Nothing with that level of scale is currently in
the realm of possibility for our use-cases, but as machine-generated
data becomes more prevalent it is going to become more of a problem.
Want to create a Jira for it and assign it to the backlog? If you have
some ideas about how this should be done or want to work on a
prototype, go for it :-)
Ethan
On Fri, Sep 2, 2011 at 6:07 PM, Jack Park<[email protected]>
wrote:
The interesting question for me is: how many records can you store in an
RDBMS before you have to think about scale up issues? I wrote a trivial
microblog in Java using Redis; I think there is a candidate fit.
Cheers
Jack
On 9/1/2011 11:53 PM, Ethan Jewett wrote:
It's certainly an interesting idea, and Lift does have some built-in
capabilities here in the form of the Record module. It would be a
pretty significant migration.
An interesting question to explore: Would this give us a performance
boost in some scenarios?
Cheers,
Ethan
On Thu, Sep 1, 2011 at 10:54 PM, Jack Park<[email protected]>
wrote:
Somewhat beyond 1.4, I'd like to ask about persistence other than
RDBMS.
For
instance, in microblogging, there is some evidence that Redis can
provide
a
scalable datastore.
Thoughts?
Cheers
Jack
On 8/29/2011 11:27 PM, Richard Hirsch wrote:
I've been thinking about the next release and wanted to start
collecting ideas for features for this release.
There are some tasks in Jira [1] but that list isn't complete:
Here are some other ideas:
* Integration with Akka / Apache Camel
* Simplification of URLs
* Change Web UI to deal with new icon
Any one else have other ideas / suggestions?
D.
[1]
https://issues.apache.org/**jira/secure/IssueNavigator.**
jspa?reset=true&jqlQuery=**project+%3D+ESME+AND+**
fixVersion+%3D+%221.4%22+AND+**status+%3D+Open+ORDER+BY+**
priority+DESC&mode=hide<https://issues.apache.org/jira/secure/IssueNavigator.jspa?reset=true&jqlQuery=project+%3D+ESME+AND+fixVersion+%3D+%221.4%22+AND+status+%3D+Open+ORDER+BY+priority+DESC&mode=hide>
