[PD-dev] [flext] Fwd: fixing Flext for 64-bit?

[Dario Pedrioli]

Hi Thomas,

I hope someone will find the time to integrate the flext-based externals
into
the autobuild structure.

Sorry but I can't: it's beyond my skills.

I've got another problem now (ubuntu 9.04 64-bit machine PD 0.42.5):

in the example buffer-1.pd (py/pyext) there's a strange behaviour:
the message [mul array3 array1 array2< works only on the first half of the
array3
instead the message [add array3 array1 array2<

the difference is in th buffer.py:

def mul(*args):
    # create buffer objects
    # as long as these variables live the underlying buffers are locked
    c = pyext.Buffer(args[0])
    a = pyext.Buffer(args[1])
    b = pyext.Buffer(args[2])

    # slicing causes Python arrays (mapped to buffers) to be created
    # note the c[:] - to assign contents you must assign to a slice of the
buffer
    c[:] = a[:]*b[:]

def add(*args):
    c = pyext.Buffer(args[0])
    a = pyext.Buffer(args[1])
    b = pyext.Buffer(args[2])

    # explicit casting to arrays is also possible
    c[:] = N.array(a,dtype=N.float32) + N.array(b,dtype=N.float32)

therefore I think that:
- the direct use of the buffer doesn't work well (printing the values give
me something like "0 0.123 0 0.234 0 0.345 etc.. " - correct values followed
by a zero)
- the casting works well

My questions are:
- am I missing some basic concept?
- this behaviour is correct (I don't think so)?
- this behaviour is due to the 64-bit?

Thanks, ciao, Dario.

PS: Sorry for the bad english.


2010/3/11 Thomas Grill <gr at grrrr.org>

> thanks for your kind words. I haven't given up the hope that someday
> someone will find the time to integrate the flext-based externals into
> the autobuild structure and, consequently, pd-extended.
> gr~~~
>
> 2010/3/9 Dario Pedrioli <dariopedrioli at gmail.com>:
> > Thanks a lot Thomas.
> >
> > Now it compile fine!
> >
> > I pick the occasion to say that your work for pd (flext, pool, py,
> etc...)
> > is fantastic. Many many many thanks!
> >
> > ciao, Dario
> >
> > 2010/3/9 Thomas Grill <gr at grrrr.org>
> >>
> >> Hi Dario,
> >> please check out py from svn and try again.
> >> gr~~~
> >> Am 09.03.2010 um 09:48 schrieb Dario Pedrioli:
> >>
> >> Hi all,
> >>
> >> I'm trying to compile py/pyext on ubuntu 9.04 64-bit machine PD 0.42.5
> >>
> >> with flext rev 3659 I've got no problems.
> >>
> >> with flext rev 3723 I've got the error below:
> >>
> >> ubuntu at ubuntu-desktop:~/pd/
> >> source/py_3723$ sh ../flext_3723/build.sh pd gcc
> >> make -f ../flext_3723/buildsys/gnumake-sub.mak  PLATFORM=lnx RTSYS=pd
> >> COMPILER=gcc BUILDPATH=../flext_3723/buildsys/ PKGINFO=package.txt
> >> BUILDCLASS=ext USRCONFIG=config.txt USRMAKE=build/gnumake-lnx-gcc.inc
> >> TARGETMODE=release TARGETTYPE=multi THREADED=1 _build_
> >> make[1]: ingresso nella directory ?/home/ubuntu/pd/source/py_3723?
> >> mkdir -p pd-linux/release-multi/
> >> g++ -c     -pthread -fPIC -fvisibility-inlines-hidden -DPY_EXPORTS
> >> -DPY_NUMPY -DPY_USE_GIL -DNDEBUG -DFLEXT_THREADS -DFLEXT_SYS=2 -DPD
> >> -I/usr/include/python2.6 -I/home/ubuntu/pd/source/pd-0.42-5/src
> >> -I/usr/local/include/stk -I/usr/local/include/sndobj
> >> -I/usr/local/include/flext source/pybuffer.cpp -o
> >> pd-linux/release-multi/pybuffer.opp
> >> source/pybuffer.cpp:154: warning: deprecated conversion from string
> >> constant
> >> to ?char*?
> >> source/pybuffer.cpp: In function ?int buffer_ass_slice(PyObject*,
> >> Py_ssize_t, Py_ssize_t, PyObject*)?:
> >> source/pybuffer.cpp:406: error: no matching function for call to
> >> ?flext_multi::CopySamples(flext_multi::buffer::Element*, t_sample*,
> int&)?
> >> /usr/local/include/flext/flsupport.h:425: note: candidates are: static
> >> void
> >> flext_multi::CopySamples(t_sample*, const t_sample*, int)
> >>
> >> Someone can help?
> >>
> >> PS: I'm trying rev 3723 because of 64-bit buffer patch (
> >>
> >>
> http://markmail.org/message/6v52tlvllwvcwilj#query:+page:1+mid:6hpplp4wfi4u6vkc+state:results
> <gr at grrrr.org>
> >> )
> >> My goal is to handle arrays directly in python, but with the rev 3659
> I'm
> >> only get half of the array inside py (the other half is set to zero). Am
> I
> >> on the right way to fix my problem?
> >>
> >> Thanks and ciao!
> >> _______________________________________________
> >> Pd-dev mailing list
> >> Pd-dev at iem.at
> >> http://lists.puredata.info/listinfo/pd-dev
> >>
> >
> >
>
>
>
> --
> Thomas Grill
> http://grrrr.org
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puredata.info/pipermail/pd-dev/attachments/20100312/82fa6500/attachment.htm>