What do you want PyPy to do? 
Open Space session at EP2007 2007-07-11
================================================
Purpose:
- collect information about needs from PyPy and 
  would-be PyPy users
- answers from PyPy developers when/where applicable

How was it done:
- everyone present wrote their primary question
  and/or interest on a post-it
- these were grouped to identify overlapping areas
  interest, largest areas were adressed first
- a summary of the questions/interests grouped per
  area can be found below and the "answers" discussed

Areas of questions/interests:

RPython: stand alone and/or interfacing external libraries
--------------------------------------------------------------
Questions/interests:
- using RPython in CPython (unless PyPy can run all my Python 
  2 code next month)
- how many are interested in RPython (in CPython, support 
  available, stability)
- RPython for Python extensions
- external library interfacing (extension modules)
- 3D graphics: Fast floating points? Fast math?

Answers/comments:
- CPython API support in PyPy is slow
- Not much attention went into improving refcounting.
- The state of RCtypes (support it, refactor and change
  direction, future at all) is unclear - needs discussions
  (how - to use ctypes at (PyPy) application level?) 
- we need strategic discussions regarding how to interface
  with external c libraries, to be discussed during sprint
- general need for refactoring - timespan will be roughly
  a year to have a PyPy that usable (while still not covering
  all extension modules)
- for stand alone RPython applications there is need for usability
  fixes (shared interest) and speed/IO extensions 
  (app-level solution)
- regarding the question on 3D graphics (fast floating points, fast
  math) the answer is that it is fairly easy to add this and there is 
  a matching interest in the PyPy dev-team in this area. 
  The need to display/visualize is then a matter of general C extension
  interfacing
    
Suggestions: 
- RCtypes: letīs experiment here and discuss approaches during the sprint

Javascript backend/related areas
-------------------------------------
Questions:
- I want to write a parser in Python and port it to javascript
- usable javascript backend
- emulate pygame on flash player

Answers/comments:
- a parser is doable but you could encounter bugs, report them
  and they will be fixed - there is a matching interest here
  in the PyPy dev-team
- there is also a document on http://codespeak.net/pypy/dist/pypy/doc/rlib.html#parsing
  regarding parser creation
- regarding emulating PyGame on a flahs player - this is doable
  but work needed, some rough estimates discussed were between
  2-4 months because itīs connected to tweaking a backend or
  creating a new backend (which is an area we know a lot about;-)
  So doable, requires work and we are not too sure about the results ;-)
  
Refactoring/usability
--------------------------------
Questions:
- how pypy goes about maintaining parts?
- SLES/Solaris package? Light production usage?
- be less  hackish to use?
- pypy 1.1.0 planning, which cleanups are we to tackle
  and who?
- fixing/improving stackless, pickling and OO-support

Answers/comments:
- this was not discussed in particular due to time constraints
- was also not prioritized because it contained several PyPy
  developers interests and we prioritized to try to answer
  "users/would-be "users" interests
- it was apparent though during the entire discussion that
  we where discussing answers in the form of "doable" and
  "refactoring needed" which is what this area is all about

Suggestions:
- general refactoring/technical direction discussion in 
  connection to the sprint
  

Other areas of interests/questions
-------------------------------------
Questions:
- reducing the memory footprint of the Python interpreter
  (I want to run preferable several, securely separated interpreters
  on a resource constrained system (a phone specifically)
- different language interpreters/front-ends

Answers/comments:
- these were not discussed due to time constraints
- they will be taken into account when discussing general
  refactoring/technical direction