Mailman 3 June 2011 - pypy-dev

Ootypes
by emailgregn＠googlemail.com 15 Jun '11

15 Jun '11

Pretty please, can ootypes get some love in the upcoming sprint? I've gotten through the cli-jit thesis and have the brainfeck tutorial compiling to C, but neither of the CLI or JVM backends work on those tutorial examples. Setting --opt=0 gets further but not all the way there. I'd like to try a parrot PIR backend but without ootypes I'm in over my head. Thank you Sent from my BlackBerry® wireless device

2 1

Win32: track_and_esp.s
by Armin Rigo 15 Jun '11

15 Jun '11

Hi Amaury, Sorry, track_and_esp.s is most probably broken on Windows due to 017e187b2716. This revision contains, among several things, a small simplification of the code because it was (probably) not possible any more to see "and esp, xxx" from gcc. So, of course, we need to partially and carefully revert this simplification... A bientôt, Armin.

2 2

libpypy.so?
by VanL 14 Jun '11

14 Jun '11

Is it possible to create an embeddable libpypy.so/dll? Are there any visible entry points into code created by the pypy toolchain? Thanks, Vna

3 5

Re: [pypy-dev] [Pyro] Anyone with Pypy experience?
by Alexandre Fayolle 14 Jun '11

14 Jun '11

On Friday 10 June 2011 20:29:53 Irmen de Jong wrote: > On 10-6-2011 9:17, Alexandre Fayolle wrote: > > On Thursday 09 June 2011 23:11:09 Irmen de Jong wrote: > >> Hi, > >> > >> I tried to run Pyro4's test suite in Pypy. > >> It didn't work very well (better on Linux than on Windows, but still). > >> > >> Is there anyone here that has experience with Pypy? > > > > I think you should ask on the pypy-dev mailing list (CC'ed) or on the > > #pypy irc channel. You will certainly get answers for the kind folks > > over there. > > Thanks Alexandre, I will probably do that once I've narrowed down the > things that didn't work. I've found and fixed 1 problem so far; Pypy > doesn't seem to like unicode keyword arguments; "TypeError: keyword cannot > be encoded to ascii". This used to be an issue with python < 2.6.? (http://bugs.python.org/issue2646). It may not have been changed in recent versions of pypy. -- Alexandre Fayolle LOGILAB, Paris (France) Formations Python, CubicWeb, Debian : http://www.logilab.fr/formations Développement logiciel sur mesure : http://www.logilab.fr/services Informatique scientifique: http://www.logilab.fr/science

3 2

static binaries please ...
by Andreas 13 Jun '11

13 Jun '11

I have just spent an hour compiling various libraries to get the pypy binaries to work (64bit, RHEL3). Probable because the distribution is old (I don't administer it) the shared libraries that pypy requires were not available. Compiling them by hand is a hassle (have to find options to enable shared objects etc.). Please distribute only _static_ binaries, the gains of shared libriares are completely insignificant compared to the hassle of when it doesn't work. Thanks ... -- -- Andreas [ http://unstable.nl | gopher://unstable.nl ]

4 4

Re: [pypy-dev] [pypy-commit] pypy default: Ah sorry, re 69cadfd7c8e1. Found how to reproduce (translate -Ojit,
by Maciej Fijalkowski 13 Jun '11

13 Jun '11

On Sat, Jun 11, 2011 at 5:53 PM, arigo <noreply(a)buildbot.pypy.org> wrote: > Author: Armin Rigo <arigo(a)tunes.org> > Branch: > Changeset: r44886:0e02f7346cf0 > Date: 2011-06-11 17:55 +0200 > http://bitbucket.org/pypy/pypy/changeset/0e02f7346cf0/ > > Log: Ah sorry, re 69cadfd7c8e1. Found how to reproduce (translate -Ojit, > of course). Add a "correct" fix. SizeDescr also uses tid. Maybe we should have BaseLLDescr simply? > > diff --git a/pypy/jit/backend/llsupport/descr.py b/pypy/jit/backend/llsupport/descr.py > --- a/pypy/jit/backend/llsupport/descr.py > +++ b/pypy/jit/backend/llsupport/descr.py > @@ -1,5 +1,6 @@ > import py > from pypy.rpython.lltypesystem import lltype, rffi, llmemory, rclass > +from pypy.rpython.lltypesystem.lloperation import llop > from pypy.jit.backend.llsupport import symbolic, support > from pypy.jit.metainterp.history import AbstractDescr, getkind, BoxInt, BoxPtr > from pypy.jit.metainterp.history import BasicFailDescr, LoopToken, BoxFloat > @@ -149,6 +150,7 @@ > > class BaseArrayDescr(AbstractDescr): > _clsname = '' > + tid = llop.combine_ushort(lltype.Signed, 0, 0) > > def get_base_size(self, translate_support_code): > basesize, _, _ = symbolic.get_array_token(_A, translate_support_code) > _______________________________________________ > pypy-commit mailing list > pypy-commit(a)python.org > http://mail.python.org/mailman/listinfo/pypy-commit >

1 0

Re: [pypy-dev] Interactive fails on pypy
by Paolo Basso 10 Jun '11

10 Jun '11

Yep... Are there plans to implement the missing functions? Thanks, P 2011/6/8 Caleb Hattingh <caleb.hattingh(a)gmail.com> > On 8 June 2011 11:15, Paolo Basso <pa.basso(a)gmail.com> wrote: > >> I tryed to run the attached code with pypy but it fails (it works >> correctly with python 2.6). >> Can anybody tell me why? Thanks. >> > > Error is this: > > 0 > Traceback (most recent call last): > File "app_main.py", line 53, in run_toplevel > File "test.py", line 25, in <module> > it.ESC_test("TEST",1.0) > File "test.py", line 14, in ESC_test > if msvcrt.kbhit(): > AttributeError: 'module' object has no attribute 'kbhit' > > Maybe this message in lib_pypy/msvcrt.py line 7 has something to do with > it: > > # XXX incomplete: implemented only functions needed by subprocess.py > >

4 13

M.Sc. student introducing myself
by Even 10 Jun '11

10 Jun '11

Hello, I'm a computer science student who intent to work on PyPy for my M.Sc. thesis. I have been a passionate Python programmer for seven years, and I additionally knows C, C++, Java, Oz, Lisp and Assembler. My academic interests are algorithms, compilers and evolutionary algorithms, and I have completed classes which covers these topics, which means I know what compilers does and why, and the optimizations they can do. I'm pleased that I have been allowed to focus my thesis on PyPy, but I don't yet know what the scope of my thesis will be. At the moment numpy support in PyPy looks interesting, and I have read the blogs/mails regarding this topic. I won't start any actual work on my thesis until the new year, but I hope to spend summer and autumn getting familiar with the project and contributing in any way I can. So far I have only read through the dev docs, and compiled PyPy from source (which took 4.6 hours with cpython and 3.7h with pypy on my laptop). I have been following this list since easter, and I can be found in #pypy on freenode. -- Even Wiik eventh(a)gmail.com

4 3

Performance challenge of mutating class variables
by David Edelsohn 10 Jun '11

10 Jun '11

I mentioned this issue in a private conversation with Carl and he suggested that I send the details to the mailinglist. The following blog post is a nice summary of a significant performance problem in IronPython's early implementation with good links to the key discussions: http://ironpython-urls.blogspot.com/2009/05/python-jython-and-ironpython.ht… Robert Smallshire implemented OWL Basic in IronPython and experienced some massive slowdowns relative to CPython and Jython. IronPython developer Dino Diehland investigated the problem and wrote an detailed blog post about the problem and the fix implemented in IronPython. The problem was due to the user program's use of a class variable as a counter, which caused "unfortunate" behavior in the IronPython implementation -- mutating the variable invalidated the cache. PyPy shows a good speedup compared with CPython already, but it is an interesting corner case. The smaller testcase created by Robert might be a good benchmark to add to PyPy Speed Center (http://www.smallshire.org.uk/sufficientlysmall/2009/05/22/ironpython-2-0-an…). - David

7 12

Re: [pypy-dev] [Pyro] Anyone with Pypy experience?
by Alexandre Fayolle 10 Jun '11

10 Jun '11

On Thursday 09 June 2011 23:11:09 Irmen de Jong wrote: > Hi, > > I tried to run Pyro4's test suite in Pypy. > It didn't work very well (better on Linux than on Windows, but still). > > Is there anyone here that has experience with Pypy? I think you should ask on the pypy-dev mailing list (CC'ed) or on the #pypy irc channel. You will certainly get answers for the kind folks over there. -- Alexandre Fayolle LOGILAB, Paris (France) Formations Python, CubicWeb, Debian : http://www.logilab.fr/formations Développement logiciel sur mesure : http://www.logilab.fr/services Informatique scientifique: http://www.logilab.fr/science

1 0