extradoc / sprintinfo / Hildes_to_Heidel.txt

planning for the time between the Hildesheim and Heidelberg sprints

clean-up areas

the following issues are meant to be 0.7 issues.

- translate_pypy and the translator class need some cleanup

- initialization of the object space is still messy

- rtyper problem: exceptions are no longer always at the end of a code
  block because of the lowlevel rewriting

- translation problem: frozen ids used as hashes are broken
  in the post-translation program

- move bits around in the annotator to make different uses
  more pluggable

- better support of math and float exceptions

- reorganizing some wrongly named things

- erasing useless files

- cleanup of import dependencies

- support for tests from external users and different platforms

- preparing the next release

- documentation about external function calls and implementing
  builtin modules

- related to the previous: documenting how interplevel marshal
  is plugged into the system

- prefixing all the C macros and names with pypy

- look into XXX issues

- support producing a windows binary, choose a suitable compiler

other issues

storing bound method on instances confuses the annotator (we can probably live with this limitation
right now, I'm not sure but it may require a large refactoring to support this)

>>> class H:
...   def h():
...    pass
>>> class C:
...   def __init__(self, func):
...      self.f = func
...   def do(self):
...     self.f()
>>> def g():
...   h = H()        
...   c = C(h.h)
>>> t=Translator(g)
>>> t.annotate([])

additional wild ideas

- thinking of an RPython flowgraph interpreter as an executable?