On Wed, Apr 22, 2009 at 2:48 PM, Maciej Fijalkowski <fij...@gmail.com> wrote:
> JS backend is translating restricted subset of Python called RPython.
> This seems to be infeasible, since JS is truly a dynamic language,

 yes.  the features of javascript are a near one-to-one match with
python.  the only things that are "truly" missing are the int / float
bugbear (which can be emulated, i realise, but in a _very_ expensive
way), read-only attributes (slots) but even that could be
[expensively] emulated, and "undefined" tends to get in the way a lot.


> that's why we removed it. The other reason why we removed it, is that
> noone seems to be interested enough in maintaining it.

 oh.  arse.

 well, i have to say: if pypy was in a position to do the same job as
pyjamas, i'd be using pypy, not pyjamas.  and would end up maintaining
it (just like i've ended up maintaining pyjamas, because i need it,
and use it).


> As far as I know pyjamas does not translate a subset of python,

 correct. we're going for as much of python2.N as we have time for,
with a view to (eventually) making pyjs a JIT python accelerator
(using V8 as the JIT engine) - just like psyco and unladen/swallow.



> but
> pythonized javascript - ie it does not preserve python features, but
> rather tries to implement python syntax which is javascript at the
> bottom.

 for web developers, it preserves as many python features in their
entirety as web developers can stand.  int/float got sacrificed,
because that really _is_ too tricky to handle / too expensive to do
"strictly".

everything else, we're going for preserving / implementing python
features as much as possible, simply because people need it.

 however there's a separate experiment - "strict mode" - which will
involve doing a python "int" class and a python "long" class etc. etc.
that implement the full semantics.

l.
_______________________________________________
pypy-dev@codespeak.net
http://codespeak.net/mailman/listinfo/pypy-dev

Reply via email to