On Mon, Oct 21, 2013 at 11:03 AM, Robert Bradshaw
wrote:
> You can iterate over a vector using "for x in ..." rather than indexing
> though.
>
Nice. I do love cython. Every time I use it for something new, I
learn that features "just work" that I never thought to try. Thanks
for your help!
-
There's not a way to construct tuples without the intermediate lists
without using the C API directly. You can iterate over a vector using
"for x in ..." rather than indexing though.
On Mon, Oct 21, 2013 at 10:42 AM, Tom Boothby wrote:
> Okay, that's almost okay... but I'm getting a list of lists
Okay, that's almost okay... but I'm getting a list of lists from the
following code -- is it possible to get a tuple of tuples instead? I
guess I can just avoid tossing the list back to the user if not.
%cython
#clang c++
from libcpp.vector cimport vector
def tupletuple(vector[vector[int]] M):
lol, really? Can I then toss that back to python?
On Mon, Oct 21, 2013 at 10:19 AM, Robert Bradshaw
wrote:
> How about
>
> cdef tupletuple(vector[vector[int]] M):
> return M
>
> On Mon, Oct 21, 2013 at 8:48 AM, Tom Boothby wrote:
>> I'm working on overhauling a class (see [1]) that
How about
cdef tupletuple(vector[vector[int]] M):
return M
On Mon, Oct 21, 2013 at 8:48 AM, Tom Boothby wrote:
> I'm working on overhauling a class (see [1]) that wraps some c++, and
> I've got just about everything working how I want... but I have a
> nagging doubt about performance
I'm working on overhauling a class (see [1]) that wraps some c++, and
I've got just about everything working how I want... but I have a
nagging doubt about performance.
Is the following fast? Can it be made fast without a bunch of ugly
python c-api stuff?
cdef tupletuple(vector[vector[int]] M):
