RE: RFC: gnu.classpath.SystemProperties

[David Holmes]

Jeroen writes:
> I considered that, but I think it would be nicer if the common
> properties stayed in the common code, but I can certainly see some value
> in making the division a little more flexible. Maybe we should have an
> additional call to VMSystemProperties at the end of the initializer, to
> allow the VM to remove or modify the properties?

Having a pre- and post- hook for VMSystemProperties seems like a good idea.

It might also be a good idea, as was recently suggested regarding the static
initialization in System itself, to factor the static initialization code
into methods so that the pre-hook could invoke them in a different order if
needed.

PS. Jeroen: the comment in my seperate email regarding setting of the alias
properties was incorrect. We don't need those set before we can create
String literals - that would be impossible as we need a string literal to
set the property. :) I need to double-check exactly how OVM had to customise
the initialization process here.

Cheers,
David Holmes