> >I have just posted to SF two patches (for review):
> >* one completes the interpreter support for nested scopes: nested scopes
> >semantic settings should be properly inherited through exec,execfile, etc .
> >It passes the usual tests.
> It also passes whatever tests I have.
> My single comment is concerning the CompilerFlags class which I feel
> should now be placed in the org.python.core package. That way it will be
> included in jar file when compiling with the --core option. I found that
> to be necessary for the 1.3 appletviewer.
Yes, it does make sense. I will move the class.
> >* the other adds nested scopes support to jythonc (see descr. and comments)
> Looks good. Obviously my tests are woefully short of any use of nested
> scopes, but there was no negative sideeffect with the jythonc patches.
I think that's the important point, when people will start using
nested scopes a lot, we will remove the eventual bugs (I hope they are
not so many <wink>). OTOH the nested scopes test suite can be compiled and
A note: jythonc continues (as before) to do partial evolution
considering just compilation order (and not the possible evaluation orders),
in the few cases where the outcome is important (base classes of a class def
etc.) this is not a problem because the former matches the normal order,
on the other hand before the patches jythonc produced errors when an undefined
value was used, this is no longer true: the proper solution (given the we
have also nested scopes) would be to add another pass to jythonc ...
but I did not feel it was worth the effort.
> I'm very impressed with the changes to jythonc. Despite the added
> complexity of nested scopes, I feel that the result have become easier
> to understand.
> Feel free to commit these two patches and close the bug reports at your
I will do that during the weekend.