Hi,

Robert Bradshaw wrote:
> On Apr 25, 2008, at 10:11 AM, Stefan Behnel wrote:
>> It would be cool if we could get Cython into compiling itself. I  
>> think that's
>> a target worth going for, and I'm close to believing that it's not  
>> far away at all.
> 
> That would be very cool, and it's something I've been thinking about  

We could even do some voodoo stuff in the setup.py file to compile Cython for
the platform by running itself uncompiled. :)

2to3? blah - Cython-to-binary rulez!


> Most of the stuff that is broken is on our todo list already.

I fixed and worked around a couple of trivial things. No need to implement
generator support for a single "yield" statement in the source, for example.

Three things remain that should be fixed in Cython:

 1) keywords differing from Python, like "signed". It would be nice if this
    could be fixed, e.g. by providing alternative interpretations of a symbol
    in the parser (not in general, just for single identifiers).

 2) "def" inside a block/function. This requires closures, as we know.

 3) getattr(a,b)/getattr(a,b,c), i.e. alternative signatures for builtins.

We should be able to work around all of them, especially 1) and 3), but the
"def" part is really worth implementing.

A fourth thing I found (and I'm not sure what to do here):

  - __file__ - this is supposed to return the file that holds the module
    *at runtime*. Really tricky to do from a C module ...

Stefan
_______________________________________________
Cython-dev mailing list
Cython-dev@codespeak.net
http://codespeak.net/mailman/listinfo/cython-dev

Reply via email to