[Python-3000] Betas today - I hope
Nick Coghlan
ncoghlan at gmail.com
Fri Jun 13 10:56:01 CEST 2008
Georg Brandl wrote:
> Nick Coghlan schrieb:
>
>> 2. Method lookup MAY bypass __getattribute__, shadowing the attribute
>> in the instance dictionary MAY have ill effects. (slots such as
>> __enter__ and __exit__ that are looked up via normal attribute lookup
>> in CPython will fit into this category)
>
> I would argue that the __enter__ and __exit__ behavior should be changed
> too.
> The reason for the current behavior is this:
>
> 2 0 LOAD_GLOBAL 0 (a)
> 3 DUP_TOP
> 4 LOAD_ATTR 1 (__exit__)
> 7 STORE_FAST 0 (_[1])
> 10 LOAD_ATTR 2 (__enter__)
> 13 CALL_FUNCTION 0
> 16 STORE_FAST 1 (_[2])
> 19 SETUP_FINALLY 18 (to 40)
>
> IOW, when "with" is compiled, the attributes are retrieved using bytecode.
> It wouldn't be hard to have a WITH_SETUP opcode (along with the already
> existing WITH_CLEANUP) that would make the bytecode read like:
>
> 2 0 LOAD_GLOBAL 0 (a)
> 3 WITH_SETUP 0 (_[1])
> 6 STORE_FAST 1 (_[2])
> 9 SETUP_FINALLY 18 (to 40)
No argument here - the PEP 343 implementation is the way it is mainly
because it involved the least messing around with the
performance-sensitive ceval loop.
That said, the with statement implementation is already a bit different
in 2.6/3.0 (it moves things around on the stack so it can avoid the
STORE_FAST/LOAD_FAST/DELETE_FAST operations):
def f():
with a:
pass
Python 2.5 disassembly:
2 0 LOAD_GLOBAL 0 (a)
3 DUP_TOP
4 LOAD_ATTR 1 (__exit__)
7 STORE_FAST 0 (_[1])
10 LOAD_ATTR 2 (__enter__)
13 CALL_FUNCTION 0
16 POP_TOP
17 SETUP_FINALLY 4 (to 24)
20 POP_BLOCK
21 LOAD_CONST 0 (None)
>> 24 LOAD_FAST 0 (_[1])
27 DELETE_FAST 0 (_[1])
30 WITH_CLEANUP
31 END_FINALLY
Python 2.6 disassembly:
2 0 LOAD_GLOBAL 0 (a)
3 DUP_TOP
4 LOAD_ATTR 1 (__exit__)
7 ROT_TWO
8 LOAD_ATTR 2 (__enter__)
11 CALL_FUNCTION 0
14 POP_TOP
15 SETUP_FINALLY 4 (to 22)
18 POP_BLOCK
19 LOAD_CONST 0 (None)
>> 22 WITH_CLEANUP
23 END_FINALLY
Cheers,
Nick.
--
Nick Coghlan | ncoghlan at gmail.com | Brisbane, Australia
---------------------------------------------------------------
http://www.boredomandlaziness.org
More information about the Python-3000
mailing list