Andrew John Hughes wrote:
2009/8/21 Kelly O'Hair <[email protected]>:
Still need reviewer ...

 6853636: Fix warnings in jdwpgen, add jdwpgen NetBeans project

 http://cr.openjdk.java.net/~ohair/openjdk7/jdk7-build-jdwpgen-6853636/webrev/

Pretty harmless code cleanup on the tool used to generate
some of the JDWP code for the debugger.

-kto




The Java part looks pretty sane to me:

  * I think the change to ConstantSetNode is particularly important,
as it's eliminating a race condition (another class could initialise
the variable after the null check but before the new).  To keep the
same lazy initialisation semantics as before, you could use the
on-demand holder idiom
(http://en.wikipedia.org/wiki/Initialization_on_demand_holder_idiom)
but I doubt a single map is going to be too much overhead to
initialise early.

I assumed the overhead wasn't an issue, correctness seemed better.

  * The change to Parse.java means exceptions will now be thrown that
were being caught before.  I can't see this being a major issue as
part of a tool, it's more of a problem in changing the semantics of
API methods but something to be aware of nonetheless.  I much prefer
the new version; swallowing exceptions on mass is very bad practice.


I also prefer to have the caught exceptions explicit, so I think this
will be better. Not that I ever expect this code to be a problem.

Sorry I can't really comment on the NetBeans changes, but it appears
to be a set of new files rather than changes; is this correct?


Just adding a netbeans project for anyone needing to work on jdwpgen.

-kto

Hope that helps,

Reply via email to