On 26 jan 2012, at 13:56, David Holmes wrote: > On 26/01/2012 10:02 PM, Rickard Bäckman wrote: >> On 01/26/2012 12:50 PM, David Holmes wrote: >>> Change looks fine, but then so did the original! Will this be handled >>> correctly by all versions of the jar command? >> >> We, have checked the contents of the version that RE is using (1.6.0_1 >> or something like that) and the one JPRT is using (1.6.0_18) and locally >> on my machine (1.7.0) I don't think I'll be able to verify _all_ >> versions of jar. It seems the "faulty" behavior of "jar" was fixed in >> 1.6.0_18 which is the first version I could find that didn't cause the >> error on the previous code. I would rather hope that we could try to use >> one and the same version across all kind of builds of the same jdk >> version. It would help to catch this kind of errors earlier... > > Sorry I meant "all versions of jar that we might build with" :) > > I think the normal RE process is to build JDK N with the FCS version of JDK > N-1, which would explain why the old JDK 6 jar is being used. JPRT has > changes to the bootjdks made "on demand". > > As long as we target both 7u and 8 we will be using two different toolsets. > But I agree that JPRT and RE should be using the same tools. That needs to be > taken up with RE and Kelly.
Ideally not just using the same tools, but they should _be_ the same systems. But I digress... > > Thanks, > David > >> Thanks >> /R >> >>> >>>> Thanks >>>> /Rickard >>
