Re: [Sqlalchemy-tickets] [sqlalchemy] #2720: new C extensions for attribute accessing

[sqlalchemy <mi...@zz...>]

#2720: new C extensions for attribute accessing
-----------------------------------+---------------------------------------
      Reporter:  zzzeek            |      Owner:  zzzeek
          Type:  enhancement       |     Status:  new
      Priority:  medium            |  Milestone:  blue sky
     Component:  cextensions       |   Severity:  very major - up to 2 days
    Resolution:                    |   Keywords:
Progress State:  not decided upon  |
-----------------------------------+---------------------------------------

Comment (by claudiofreire):

 The savings are in the avoidance of creating and initializing python
 frames, which is the main source of function call overhead in python. Any
 function call in python will initialize a frame, some will also create one
 (there's a free list to avoid that), so that's why it cannot be done in
 python.

 Attribute access in python is very very fast, so the overhead of
 initializing a frame for such access is noticeable.

 The caching is there only as a tool to prevent calling python code (which,
 again, pays the full cost of initializing a frame, no matter how fast that
 function is inside).

 I could find no way to achieve that in pure python. Usually, I use
 operator.attrgetter to create fast attribute-getting functions, but this
 descriptor does a little bit more than just accessing the attribute, so
 that's not viable here.

--
Ticket URL: <http://www.sqlalchemy.org/trac/ticket/2720#comment:26>
sqlalchemy <http://www.sqlalchemy.org/>
The Database Toolkit for Python