Hello everyone, This is another update about our attempt to replace Jackrabbit with ModeShape in Sling.
First of all, sorry for the long wait, but we needed to be sure we had the OK from our customer before releasing what we have done in this period. The good news is that we are ready to release our work, you can find the projects here: https://github.com/sourcesense/stone and here: https://github.com/sourcesense/modeshape-stone The second one is basically a fork of modeshape with some patches we still have to submit, but at the moment we're unsure if these are "real" fixes or not; maybe some feedback from you will help to clarify this matter. The bad news is that the project, in the interest of releasing early and releasing often, is not fully working yet. We have two blocking issues, with two different configurations, and you can read more about them in the 'Issues' section of the project site. Obviously every feedback (and contribution) will be well accepted, and for the ones interested, please don't hesitate to ask also using our internal mail alias [email protected] Thanks for your patience and understanding Ale On Feb 15, 2011, at 12:02 PM, Felix Meschberger wrote: > Hi > > thanks for the feedback and congratulations to your success. > > Along with Mike Moulton and Mike Müller I invite to contribute back > either in the form of contributing the project setup or with some > documentation on our wiki. > > Thanks ! > > Regards > Felix > > Am Mittwoch, den 09.02.2011, 15:54 +0000 schrieb Alessandro Novarini: >> Hello list, >> >> As promised, here a quick update and a not-so-quick question. >> >> I think we're almost done with the modeshape-server bundle, we fought a lot >> with the class loading but now it looks good. >> >> But... >> >> We started this work assuming that replacing the jackrabbit-server bundle >> was enough to have a sling working, as Carl also confirmed right below. >> Probably we still miss something, but we are now struggling with >> authentication and authorization; in particular two issues: >> >> ModeShape relies on JAAS to check for valid credentials, and at the moment >> it's complaining that no configuration is found. >> We tried to add a jaas config file, made Sling point to that file, but we >> don't know if there are LoginModules already available or if we need to add >> one to the classpath (we asked to ModeShape forum and they suggested to look >> at JBoss Picketbox [1]). >> >> So at the moment we replaced the mechanism of authentication with a custom >> one, and ModeShape is quite happy with that, even though we would like to >> align the functionality with what Sling is using and is expecting by >> jackrabbit. >> We came across a strange side-effect there, because now from the web console >> we cannot do anything without having Sling asking for a login. >> Useless to say that whatever we enter, the login fails :( >> >> Going a step further... we're examining the class >> org.apache.sling.extensions.webconsolesecurityprovider.internal.SlingWebConsoleSecurityProvider, >> the one that tries to login to the repository passing a SimpleCredentials >> (triggering the JAAS mechanism skipping our custom one) - method >> authenticate(String, String). >> >> What scared us was the code just after the repository.login(creds): it seems >> a bit implementation-dependent, since all the code is into an if (session >> instanceof JackrabbitSession), otherwise the method will return null (no >> success). >> >> Do you confirm what we are thinking, that even fixing our current issue, >> we're going to have a "broken" web console? >> >> Thank you in advance >> Ale >> >> >> On Jan 13, 2011, at 4:11 AM, Carl Hall wrote: >> >>> We (Sakai Project) are working to replace JCR with a non-JCR backend >>> (Cassandra and any JDBC; our mail list details the reasons and factors). >>> While this isn't JackRabbit -> ModeShape (staying with JCR will be much >>> easier), we can certainly attest that changing the storage engine is >>> possible with some work. The Sling code does a great job of decoupling >>> itself from the storage mechanism. >>> >>> >>> On Fri, Dec 17, 2010 at 10:27 AM, Alessandro Novarini < >>> [email protected]> wrote: >>> >>>> Hi Simon, >>>> >>>> Fine for me, we should start working on the server bundle soon. >>>> I'll let you know what our progresses are. >>>> >>>> Thanks >>>> Ale >>>> >>>> >>>> On Dec 17, 2010, at 2:28 PM, Gunzenreiner Simon wrote: >>>> >>>>> Hi Ale >>>>> >>>>> I posted this question originally. Just to give you an update on where we >>>> stand - we haven't started this work yet, and it probably will be mid >>>> January until we do. I suggest we share our findings on this list ... >>>>> >>>>> Regards >>>>> Simon >>>>> >>>>> -----Ursprüngliche Nachricht----- >>>>> Von: Alessandro Novarini [mailto:[email protected]] >>>>> Gesendet: Mittwoch, 15. Dezember 2010 18:43 >>>>> An: [email protected] >>>>> Betreff: Sling+ModeShape >>>>> >>>>> Hello list, >>>>> >>>>> I've just subscribed because I would like to know if any progress has >>>> been made on the replacement of Jackrabbit with another JCR implementation, >>>> as discussed here: >>>>> >>>> http://dev.day.com/discussion-groups/content/lists/sling-users/2010-11/2010-11-16_Replacing_jackrabbit_with_other_JCR_implementation_Gunzenreiner_Simon.html >>>>> >>>>> I've tried in the last couple of days to work on it too, taking as sample >>>> the org.apache.sling.auth.openid bundle, and I've had some hard time >>>> because >>>> at the moment I have little knowledge of ModeShape; anyway, if there was >>>> the >>>> will to work openly on the project, I would be glad to contribute in some >>>> way. >>>>> >>>>> I don't have the original thread as email so sorry for not directly >>>> replying to it. >>>>> >>>>> Thanks >>>>> Ale >>>> >>>> >> > >
