On Thu, 12 Oct 2006 10:36:53 -0700, Justin C. Walker <[EMAIL PROTECTED]>
wrote:
>> This "final word" might take a while: omalloc is not a shared
>> library and thus
>> I either have to make it one or link it in statically which
>> probably won't
>> work at all (because this would result in sever
On Thursday 12 October 2006 17:36, Justin C. Walker wrote:
> On Oct 12, 2006, at 09:34 , Martin Albrecht wrote:
> >> Guys, this is not where the bulk of our time is at the moment. The
> >> bulk of our time now is in constructing the *Python* object, not in
> >> constructing or managing the underly
On Oct 12, 2006, at 09:34 , Martin Albrecht wrote:
>
>> Guys, this is not where the bulk of our time is at the moment. The
>> bulk of our time now is in constructing the *Python* object, not in
>> constructing or managing the underlying GMP object. The python
>> integer type manages to avoid con
On Thu, 12 Oct 2006 09:34:34 -0700, Martin Albrecht
<[EMAIL PROTECTED]> wrote:
>> Guys, this is not where the bulk of our time is at the moment. The
>> bulk of our time now is in constructing the *Python* object, not in
>> constructing or managing the underlying GMP object. The python
>> integer
> Guys, this is not where the bulk of our time is at the moment. The
> bulk of our time now is in constructing the *Python* object, not in
> constructing or managing the underlying GMP object. The python
> integer type manages to avoid constructing Python objects altogether
> in most cases (it reu
I was also suggesting that sage do it transparently. That is the
user is not expected to identify the intermediate object and write
code for it, but that sage translates it to the code with reusable
intermediate objects.
Rishi
On 10/12/06, William Stein <[EMAIL PROTECTED]> wrote:
>
> There have
There have been proposals for mutable integers, and that's what you're
suggesting below. I think there should be a class
MutableInteger
that derives from integer and is mutable. These would offer some
ops like you suggest below.
-- William
On Thu, 12 Oct 2006 08:06:50 -0700, R Rishik
It may be possible to minimize the intermediate object creation
by introducing intermediate objects at the beginning of a block
and reusing them (especially in loops).
example
for i in range(100):
f(a+i)
is translated to
c=0
for i in range(100):
c.set_to_sum(a,i)
This reduce
On Oct 12, 2006, at 8:39 AM, [EMAIL PROTECTED] wrote:
>
> On Thursday 12 October 2006 02:55, Craig Citro wrote:
>> As far as speeding up arithmetic
>> with integers, there's one "known" trick that people use, namely
>> switching
>> between boxed & unboxed integers.
>
> Yeah, this was my sugges
On Thursday 12 October 2006 02:55, Craig Citro wrote:
> As far as speeding up arithmetic
> with integers, there's one "known" trick that people use, namely switching
> between boxed & unboxed integers.
Yeah, this was my suggestion as well. It seems to me that it's a hands-down
win if you mostly
10 matches
Mail list logo
