Re: [Swig-user] =?utf-8?q?C++_STL_for_python=3A_iterators=2C_functors?= =?utf-8?q?=2C_algorithms=2C=09and_of_course_containers?=

[William S F. <ws...@fu...>]

Eric Mahurin <eric.mahurin <at> gmail.com> writes:

> 
> 
> On 5/1/07, Nitro <nitro <at> dr-code.org> wrote:
> 
> 
> Am 02.05.2007, 01:23 Uhr, schrieb Eric Mahurin <eric.mahurin <at> gmail.com>:>
I have a series of hacks to the swig 1.3.31 python .i and .swg files that> gives
the following STL functionality:
> >> * containers (vector, list, deque, set, map, multiset, multimap) of>
python> objects that even use custom python compare functions.>> * read/write
random-access iterators in those containers.
> >> * functors from callable python objects (i.e. lambdas, functions,>
methods).> These can operator on normal python objects or those that swig
wrapped> (i.e.> C++ objects in a container).
> >> * ability to easily wrap most(all?) algorithms to operate on iterators>
from> any container (and use the above functors).
> 
>  
> 
> Sounds really cool. I guess you are the official swig python stl expertnow 
This could probably even be ported to be part of the UTL (so thatnobody except
you will be able to read the code ;) ).Joking aside, in my opinion these things
should go in, probably along with
> some examples (if you can be bothered with that) and/or tests
anddocumentation. Withouts docs it's one of those features you'll never findout
about unless you dig five hours through the swig source and it'll take
> you another five hours to understand it.Of course William is the one who
decides if this should go in. He'll alsotell you the best way to get it into SVN
(probably a patch or the fullfiles or something).
> -Matthias
> 
> 
> I've only been using swig for a little less than a week, so I can't be that
much of an expert.  I mainly was just writing the C++ code necessary to get
these things to work.
> I posted what I have
here:http://sourceforge.net/tracker/index.php?func=detail&aid=1712193&group_id=1645&atid=301645
> It includes the .swg and .i files necessary, the same description from this
post, and a little bit of python testing.  The full versions (not patches) of
the hacked .swg and .i files are there, so you don't need to put these in the
swig install since it searches in the current dir first.

Submitting a patch on SF is the best approach to getting it into the
distribution. I'm afraid that I'm not familiar enough with the UTL to commit
these patches myself, but any of the SWIG developers who is comfortable with
this are welcome to commit and test out all the languages that depend on it.

With regard to iterators, you might want to take a look at the Ruby improvements
in this area that Gonzalo has committed recently.

William