>> I dont have any experience with Thrift+Cayenne, but I'd be happy to include >> that configuration/technology stack in any approach.
> Bryan can probably help there I imagine..? In fact I have no experience with Cayenne. Of the projects listed so far, I'm most probably comfortable with Cassandra (twissandra project is an excellent crash-course in that btw: http://github.com/ericflo/twissandra) I've done projects which had their own database abstraction (one had to support Derby, MySQL, PostgreSQL, DB2, Oracle all at the same time with connection pooling). This was before tools like Cayenne were available... not to mention, a naiive approach using .properties files and generous switch statements to manage pools were sufficient for my purposes, but I'd hazard to guess not so much in an Apache community-quality full release. Can't promise any brilliant code or anything, but I'll at least look into Cayenne and compare with what I already know about Hibernate, JPA and no-SQL tools so that I might be able to at least make a more reasonable recommendation, if not contribute some sample code at some point of how to connect Wookie to at least one of the no-SQL datastores... Bryan -----Original Message----- From: Scott Wilson [mailto:[email protected]] Sent: May 11, 2010 3:04 PM To: [email protected] Subject: Re: important todo: remove hibernate On 11 May 2010, at 18:53, Randy Watler wrote: > Scott: > > I am willing to take on this project starting immediately if it can help you > guys and I would not be getting in the way of the ongoing work of others. > Like I said, I have to support JCR storage for Wookie in any case ASAP. That sounds great to me - what do other committers think? I can imagine other implementations where a JCR connection is going to be useful (Sakai comes to mind, for example). > OJB is what Jetspeed uses internally... I would suggest using JPA unless > there is a compelling reason to do otherwise. I only mention OJB as a plugin > candidate that Jetspeed might use in the future, not an important one that > Wookie support natively. OK, that makes sense. JPA, JDO, EmpireDB etc are all in the same category, as relatively straight switches for HIbernate. > I dont have any experience with Thrift+Cayenne, but I'd be happy to include > that configuration/technology stack in any approach. Bryan can probably help there I imagine..? > > Let me know, > > Randy > > Scott Wilson wrote: >> On 11 May 2010, at 18:22, Randy Watler wrote: >> >> >>> Kris/All: >>> >>> Hello... I am currently starting to modify Wookie to use a JCR backend for >>> use in a CMS web site environment. I am also interested in making the >>> solutions pluggable along the way so that other implementations can be used >>> to suit the environment. I am a committer on the Jetspeed project and there >>> is interest there as well, so using the native store there, (OJB), would be >>> ideal. Obviously, this thread has mentioned other candidates! >>> >>> How best can I help you guys here make this happen? >>> >> >> Hi Randy, >> >> The most pluggable we can be the better I reckon. So things like JCR, JPA >> and so on do have an advantage in that they allow multiple implementations. >> However if we can also make it possible to have a Thrift connection to >> Cayenne then that's good too! >> >> I put OJB on the candidate list, but when I looked at the I couldn't see >> much activity (last commit back in 2008) - though maybe thats just because >> its mature. What's your experience of OJB in Jetspeed? >> >> - Scott >> . >>> Randy >>> >>> Kris Popat wrote: >>> >>>> On 11 May 2010, at 15:07, Copeland, Bryan wrote: >>>> >>>> >>>>> Kris, >>>>> >>>>> When you mentioned building your own file-based solution it made me think >>>>> of the growing "no-SQL" movement. I wonder if it might be useful to >>>>> leverage yet again another Apache project, Cassandra: >>>>> http://cassandra.apache.org/ >>>>> >>>>> For "very large-scale" systems (and likely much more complicated), is >>>>> Apache Hadoop: >>>>> http://hadoop.apache.org/ >>>>> >>>>> Probably most know about these, but they are examples of key-based and >>>>> graph-based storage systems, respectively... Document-based approaches >>>>> already exist too, so it may make sense to leverage the work done by the >>>>> CouchDB team: >>>>> http://couchdb.apache.org/ >>>>> >>>>> Just thought I'd share that as the first two projects, and Wookie, are >>>>> currently the three up and coming Apache projects my organization is >>>>> tracking most closely. I'm not sure who will win the >>>>> "efficiency/lightweight data store war", but in the end, an approach >>>>> which offers options and flexibility for datastore configuration will >>>>> probably be the nicest for the community, but, most difficult to >>>>> accomplish because of the differences between RDBMS and Graph-based >>>>> camps, perhaps Document-based might be a nice middle-ground though? >>>>> >>>> Thanks for that, will add them to the list of possibilities on the wiki. >>>> These look very interesting. Best for us is to find something that slots >>>> in easily replacing the current db middleware that we are using taking >>>> issues of robustness, extensibility, load handling and licensing into >>>> consideration. Will be spending some time on this over the next few days >>>> tinkering and evaluating >>>> >>>> >>>> >>>>> Bryan >>>>> >>>>> >>>>> -----Original Message----- >>>>> From: Kris Popat [mailto:[email protected]] >>>>> Sent: May 11, 2010 5:28 AM >>>>> To: [email protected] >>>>> Subject: Re: important todo: remove hibernate (was Re: Fwd: Several >>>>> podling reports still missing at >>>>> http://wiki.apache.org/incubator/May2010, due today) >>>>> >>>>> >>>>> On 11 May 2010, at 09:21, Scott Wilson wrote: >>>>> >>>>> >>>>>> On 11 May 2010, at 09:16, Ross Gardler wrote: >>>>>> >>>>>> >>>>>>> On 11/05/2010 09:05, Scott Wilson wrote: >>>>>>> >>>>>>>> I've updated the report here: >>>>>>>> >>>>>>>> http://wiki.apache.org/incubator/May2010#Wookie >>>>>>>> >>>>>>> +1 to your report. >>>>>>> >>>>>>> A busy quarter. >>>>>>> >>>>>>> I'm largely silent at present due to spending all my time on >>>>>>> http://www.transfersummit.com >>>>>>> (people should come, it's a great conference). >>>>>>> >>>>>>> Once that's out of the way I want to really crack on with getting >>>>>>> rid of hibernate so we can get a release out the door. In my >>>>>>> opinion, we need a release to really start building community. >>>>>>> >>>>>>> Of course, if someone wants to get cracking on that before me I'll >>>>>>> gladly start a branch for that work and keep it aligned with trunk >>>>>>> for you (asuuming you are not already a committer). >>>>>>> >>>>>> Yes, that's pretty much the last hurdle. >>>>>> >>>>>> Kris, were you going to put together a list of candidates for >>>>>> replacing Hibernate on the wiki? >>>>>> >>>>> Yes I've looked through some options a couple of weeks ago, will pick >>>>> it up again and put some ideas up. >>>>> >>>>> It might be worth testing a file based solution that I've been working >>>>> on too. I'll put a patch up for people to test in a few days time. >>>>> Will need testing for robustness and speed. >>>>> >>>>> >>>>>>> Ross >>>>>>> >>>> Kris >>>> >>>> >>>> >> >> >> >
