On 10/12/2015 20:44, serguei.spit...@oracle.com wrote:
:
3. In the Module command set then we'll need to decide the reply to
the Name command for the case that the module is an unnamed module.
There is also an open issue for the runtime API too.
The empty string is returned in the implementation.
Would it be Ok to update the jdwp spec with this?
We haven't decided what Module::getName should return for unnamed
modules yet, it currently returns null. So we'll have to come back to
it. For now then maybe JDWP can just return the stringID as 0 and we'll
come back to it.
:
In ReferenceTypeImpl then I assume isModuleCached is not needed.
Not sure, I understand this. Why?
It seems there is some confusion here.
This flag is similar to the flag isClassLoaderCached.
I should have been clearer, I was just making the point that the flag
isn't really needed as module == null will tell you that it hasn't been
retreived from the target VM.
Have you thought about SA yet? I can't recall if it is compiled with
the boot JDK or will be compiled against the newly built jdk.jdi
module. If the later then I assume that SA will need updates. If the
former then I assume we will have issues with boot cycle builds.
I'll ask Dmitry as he covers the SA.
He had some plans on the Jigsaw update.
Okay, although I'm just asking about the SA JDI connectors here. We just
need to make sure that they build and I can't recall if they are
compiled against the JDI in the boot JDK or the jdk.jdi module. If the
former then we would only see issues with boot cycle builds.
-Alan.
