Alex lets back up here -- we are using deployment model 1 (i think) and use the jcr api directly in the code. every request generates a transientrepository. this is all wired together with spring, which is configured with the location of the cnd file and the repo.xml.
so. every time a new request comes in a TR is generated which lives as long as the session. I don't know if spring is reusing sessions or not. since the sessionfactory, which is attached to the repository factory, is configued with the cnd file and repo xml, what is happing with the filesystem to support these repositorys? In any event, the reason for "sudden" test failures was found; another test (not transactional) was leaving artifacts in the db that prevented the dao tests from running. So the claim that they worked fine for a period of time and started failing randomly should have been "worked fine until the locking tests were run, which left artifacts in the repo which prevented the dao tests from running". On Fri, Aug 22, 2008 at 3:19 PM, Alexander Klimetschek <[EMAIL PROTECTED]>wrote: > On Thu, Aug 21, 2008 at 3:43 PM, Michael Harris > <[EMAIL PROTECTED]> wrote: > > So the next time the app starts up, the cnd files are not copied into > memory > > and not used by the repo to decipher the node types? > > Things (cnd, namespaces) are stored *in* the (Memory)FileSystem. If > that store is gone, there is no other place to get them. > > Regards, > Alex > > -- > Alexander Klimetschek > [EMAIL PROTECTED] > -- --------------------- Michael Harris
