Re: Memory residence

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Per" == Per Bothner <per@bothner.com> writes:

Per> The hope is that the compiler
Per> will generate a compressed mostly-pointer-free representation,
Per> which can be expanded lazily at run-time when needed.

Per> However, I don't believe we have a concrete design or timeline
Per> for this.

Yeah, that sums it up.  One idea Andrew had was that we could make it
a slight variant on the .class format and re-use most of the existing
class file parser.

We have noticed that the GC shows up rather high when oprofiling.  If
this change would have a significant impact, that would probably push
it higher up the priority list.

Tom