On Jul 13, 2010, at 11:21 AM, David Blevins wrote: > > On Jul 13, 2010, at 9:48 AM, David Jencks wrote: > >> >> On Jul 13, 2010, at 8:13 AM, Jean-Louis MONTEIRO wrote: >> >>> >>> Hey guys, >>> >>> Most of our CI build fail. >>> IMHO, it's not relevant to use a CI system if we never take build results >>> into account. >>> Some users noticed that too, and we should fix that situation. >>> >>> So, I tried digging into continuum logs and fix all issues (as much as >>> possible). >>> To name but a few, i dug into an XMLInputFactory issue (thanks Thiago) >>> recently introduced in the trunk. >>> >>> http://download.oracle.com/docs/cd/E17477_01/javaee/5/api/javax/xml/stream/XMLInputFactory.html >>> http://download.oracle.com/docs/cd/E17409_01/javase/6/docs/api/javax/xml/stream/XMLInputFactory.html >>> >>> As described by the javadoc, #newFactory should be available in Java 6 but >>> actually, it appears only in a minor release. >>> >>> Anyway, I registered to Continuum in order to edit OpenEJB jobs. I was >>> looking if we could easily switch to the latest VM. >> >> For this particular issue maybe we should be compiling/using the geronimo >> stax api and woodstox stax implementations? Not sure how hard this is to >> set up in maven. There should be a couple other places where we need >> something more advanced than the jvm, such as annotations 1.1. > > Maybe we could offer a fallback way to parse the xml if they don't have that > particular class available.
well, its part of stax api 1.2 which is part of the ee6 platform, so it ought to be available. On the other hand setting maven up to really use dependencies to replace stuff in the jdk can be tricky. Not sure how hard it would be in this case.... but it's just a test and there's an equivalent stax 1.0 api, so I changed the code to use that one.... which ought to be available wherever stax is used. thanks david jencks > > > -David >
