On Saturday, 21 April 2012 09:25:40 UTC+1, Steven D'Aprano wrote:
> On Fri, 20 Apr 2012 09:10:15 -0700, Jon Clements wrote:
>
> >> But I don't know how. I know that I can see the default arguments of
> >> the original function using func.__defaults__, but without knowing the
> >> number and names of func's positional arguments (which I don't know how
> >> to find out) this doesn't help me. Any suggestions?
> >
> > Possibly take a look at functools.lru_cache (which is Python 3.2+), and
> > use the code from that (at it's part of the stdlib, someone must have
> > done design and testing on it!).
>
> With respect Jon, did you read the Original Poster's question closely?
> Using a LRU cache doesn't even come close to fixing his problem, which
> occurs *before* you do the lookup in the cache.
I did indeed Steven - what I was suggesting was that functools.lru_cache would
be a good starting point. Although I will completely admit that I didn't read
the code for functools.lru_cache thoroughly enough to realise it wouldn't be
suitable for the OP (ie, it sounded right, looked okay at a glance, and I
figured it wasn't a totally unreasonable assumption of a suggestion - so guess
I fell into the old 'assume' trap! [not the first time, and won't be the last
for sure!])
>
> Rotwang's problem is that if you have a function with default arguments:
>
> def func(spam=42):
> return result_of_time_consuming_calculation()
>
> then these three function calls are identical and should (but don't)
> share a single cache entry:
>
> func()
> func(42)
> func(spam=42)
>
> The OP would like all three to share a single cache entry without needing
> two redundant calculations, which take a long time.
>
> The problem is that the three calls give three different patterns of args
> and kwargs:
>
> (), {}
> (42,) {}
> (), {'spam': 42}
>
> hence three different cache entries, two of which are unnecessary.
I'm wondering if it wouldn't be unreasonable for lru_cache to handle this.
Cheers, Jon.
--
http://mail.python.org/mailman/listinfo/python-list
