On Sun, Mar 18, 2012 at 5:22 PM, Roman Shaposhnik <r...@apache.org> wrote: > On Thu, Mar 15, 2012 at 12:46 PM, Jos Backus <j...@catnook.com> wrote: >> "Finally, Red Hat seems to have largely switched to OpenJDK. This is >> what JPackage is using for builds as well. > > Speaking of which -- what's your read on JPackage. It feels dead to me > (and that's why Bigtop started packaging outside of immediate Bigdata > scope) but I could be wrong. > >> There has been no need to >> build a nosrc.rpm for a while now, assuming you're willing to give up >> the Oracle JDK for the OpenJDK in your base OS." >> >> How do people here feel about this move? > > I think we have to bite the bullet and start supporting OpenJDK. At least > as an alternative JDK to Oracle's one.
Various folks have been trying out Hadoop on OpenJDK with Whirr (see https://issues.apache.org/jira/browse/WHIRR-518), and no major problems have been found yet. Granted, it hasn't had extensive testing, but it's a start. Doing some testing in Bigtop would be great. > That said I had reports about > silly incompatibilities (like what a null would produce when turned into > a string or something) that trip up Hadoop big time and nobody wants > to spend time debugging it. Can you point to any bug reports please? Tom > > Thanks, > Roman.
