[pypy-dev] How's the JIT coming along?

Erik Gorset erik at medallia.com
Wed Aug 22 17:47:20 CEST 2007


On Aug 22, 2007, at 4:13 PM, Carl Friedrich Bolz wrote:
> those inline checks are not enough in the Python case. You also need
> to check that the method A.foo was not changed in the meantime. But
> yes, in principle this is the idea. It works even better in the case
> of integer-handling functions, because the inlined operations can in
> many cases be implemented by processor opcodes.

Yes, and one way to do this is to keep a dependency graph for the  
compiled
code, and invalidate when a dependency changes. You can get away with
checks at relatively few points, and you can also insert traps for  
assignments
to objects which will (possibly) shadow compiled dependencies.

-- 
Erik Gorset

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 1962 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/pypy-dev/attachments/20070822/394ad402/attachment.bin>


More information about the Pypy-dev mailing list