Mailman 3 March 2008 - pypy-dev

Next sprints
by Armin Rigo April 4, 2008

April 4, 2008

Hi all, This e-mail is about starting the discussion for ideas and plans about the next sprints. What we have planned at the moment: * The next PyPy sprint is going to be a small one at PyCon (http://us.pycon.org/2008/sprints/projects/#pypy). * In Leysin, beginning of April, will probably be held a "Paper writing" sprint. Confirmation and more details should come soon (from Michael Leuschel and Carl Friedrich). Another idea I now about is a follow-up Smalltalk-in-RPython (SPy) sprint (nothing concrete though). It seems that we could starting planning for a regular PyPy development sprint soon. I'm never tired of inviting people to Leysin but I guess we could pick another location now (additionally, there are many places where Spring is a nice season, but Leysin is not one of them). A bientot, Armin

3 3

Re: [pypy-dev] [pypy-svn] r53194 - pypy/branch/jit-hotpath/pypy/tool
by Maciek Fijalkowski March 31, 2008

March 31, 2008

arigo(a)codespeak.net wrote: > Author: arigo > Date: Mon Mar 31 10:51:44 2008 > New Revision: 53194 > > Modified: > pypy/branch/jit-hotpath/pypy/tool/udir.py > Log: > This logic seems slightly broken - I end up with tons of different > usession names in my /tmp, not just branch names. Anyway I prefer > the old behavior by far, so with this checkin I can set > PYPY_USESSION_BASENAME to a simple '-' and get it back. > > > Modified: pypy/branch/jit-hotpath/pypy/tool/udir.py > ============================================================================== > --- pypy/branch/jit-hotpath/pypy/tool/udir.py (original) > +++ pypy/branch/jit-hotpath/pypy/tool/udir.py Mon Mar 31 10:51:44 2008 > @@ -17,9 +17,11 @@ > else: > return basename.split('/')[-2] > > -try: > - basename = '-' + svn_info(py.path.svnwc(py.magic.autopath().dirpath()).info().url) + '-' > -except: > - basename = '-' > +basename = os.environ.get('PYPY_USESSION_BASENAME') > +if not basename: > + try: > + basename = '-' + svn_info(py.path.svnwc(py.magic.autopath().dirpath()).info().url) + '-' > + except: > + basename = '-' > > udir = local.make_numbered_dir(prefix='usession' + basename, keep=3) > _______________________________________________ > pypy-svn mailing list > pypy-svn(a)codespeak.net > http://codespeak.net/mailman/listinfo/pypy-svn > > :. > > That's because it contained tons of bugs and over checkins you got different dirs :-( I hope it's fixed by now :.

1 0

PyPy GSoC Application: Supporting Python 2.5 features in PyPy
by Bruno Gola March 29, 2008

March 29, 2008

Hi, I'm a under-grad student (3th year) from "Centro Universitário SENAC" from São Palo, Brazil. I'm very interested in the PyPy project and since last year I have been planning to get involved. Well, this plans has transformed into my graduation project that I'll start working next semester. My graduation project idea is to study two different implementations (CPython (C/Static/lower-level) and Pypy (Python/Dynamic/higher-level)) of a programming language (Python) and compare the difficult to hack both. I intend to study both and suggest optimizations to the implementations and show that it is easier to modify the higher-level implementation. Well, I told you this because my GSoC application is all about the Pypy interpreter (or the Pypy implementation of Python ) ;). I wrote some ideas to the project, so I ask if anyone could review it. PROPOSAL: Support new features of Python 2.5 in PyPy ========================================== * Motivation and goal Pypy's interpreter currently supports some of the "new" Python features introduced in Python 2.5. Also, the Python 2.5 changes to standard library have not been ported yet. This proposal goal is to bring 2.5 features and changes to Pypy interpreter. * Features Some features are already supported (for example __index__). I'm proposing to make Pypy support all the remaining features from 2.5. There are changes made to the python standard library[1] that haven't been ported yet because of the lack of 2.5 features, so I'm proposing to port this changes (some major modules added are sqlite3, ctypes and ElementTree) too. So, what I'm proposing: * Support all Python 2.5 features * "Sync" PyPy's standard library with CPython [1] http://docs.python.org/whatsnew/modules.html * Coding I started looking into the list of what is new in Python 2.5 to see if any new feature would need a special attention. My plans are to start coding the changes to the language (not to the stdlib) because I think it doesn't make much sense to implement the stdlib without full language support and it seems that porting the stdlib will not be much trouble (not as much as supporting some features :-)). I intend to start coding some simple feature (like PEP-342 that seems easy to support) so I can study the PyPy interpreter better. Then I'll code the other features (PEPs 314[is this really needed in Pypy?]], 328, 342, 343[or what is missing] and 352). [am I missing any other feature?]. With all features supported (and tested :-)) I'll code the stdlib changes. At this point I think I'll be familiar with the interpreter code so I pretend to port first the sqlite3 module. I like follow TDD (Test-driven development) so tests will not be a problem (I mean, it will be part of the final code, of course :)). PROPOSAL END. Thanks, -- Bruno Fialho Marques Gola <brunogola(a)gmail.com> bgola` @ irc.freenode.org http://www.brunogola.com.br Cel: (11) 9294-5883

2 2

Re: [pypy-dev] Looks like Macij's talk was a hit with the attendees
by Laura Creighton March 29, 2008

March 29, 2008

And looks like I cannot spell this morning, sorry about that Maciej Laura

1 0

Looks like Macij's talk was a hit with the attendees
by Laura Creighton March 29, 2008

March 29, 2008

Alas, I still cannot get them to remove my name from the list of presenters. Forward from Python-organisers From: Brian Ray <bray(a)sent.com> To: pycon-organizers Pycon-Organizers <Pycon-organizers(a)python.org> Since we are nearing the 50% participation mark on survey collection, I think it is fair to start analyzing the results. Once the paper forms are hand entered, I am happy to say, we will pass 50% user participation in the survey. This alone, speaks volumes. Here is a link, if you prefer raw data view: <http://tinyurl.com/2eoab2> <snip> OVERVIEW ======== On the highest level, here are some comparisons between the last year data (found here <http://wiki.python.org/moin/PyCon2007/Feedback>): - 500+ Surveys were filled out this time, versus 150, last time - There were slightly more first timers this year 60.6% than last year 56% - Overall satisfaction dropped slightly from 3.62/4 to 3.28/4 - Satisfaction with keynotes dropped from 3.28/4 to 2.78/4 - Satisfaction with talks dropped from 3.12/4 to 2.81/4 - Satisfaction with the network dropped sharply from 3.01/4 to 2.19/4 - Satisfaction with the food dropped from 3.0/4 to 2.61/4 - Likelihood of attending next year dropped from 3.4/4 to 3.26/4 - There was an increase of attendance of tutorials from 31.1% to 44.4% The results on preference of Only weekdays, including one weekend days, and including two weekend days, remained about the same. The top 5 talks (including keynotes) were: - Keynote: Guido van Rossum 41.4% 173 - Core Python Containers -- Under the Hood Mr. Raymond D Hettinger 20.1% 84 - Keynote: Brian 'Fitz' Fitzpatrick 18.9% 79 - The State of Django Adrian Holovaty 17.7% 74 - Keynote: Van Lindberg 16.3% 68 The top 10 talks not including keynotes: - Core Python Containers -- Under the Hood Mr. Raymond D Hettinger 20.1% 84 - The State of Django Adrian Holovaty 17.7% 74 - Sights and sounds with pyglet Mr. Alex Holkner 14.8% 62 - The State of PyPy Maciej Fijalkowski (merlinux GmbH) ; Ms. Laura A Creighton (Open End AB) 13.6% 57 - SQLAlchemy 0.4 and Beyond Mike Bayer 13.2% 55 - Using PyGame and PySight to Create an Interactive Halloween Activity Mr. John Harrison (Insight Industries) 12.9% 54 - Consuming HTML Ian Bicking (The Open Planning Project) 11.2% 47 - nose: testing for the lazy coder Jason Pellerin (Leapfrog Online) 10.5% 44 - IronPython: The Road Ahead Mr. Jim Hugunin (Microsoft) 10.5% 44 - Tahoe: A Robust Distributed Secure Filesystem Brian Warner 10.3% 43 ------- End of Forwarded Message

1 0

Re: [pypy-dev] JIT SoC
by Laura Creighton March 28, 2008

March 28, 2008

In a message of Fri, 28 Mar 2008 15:42:53 +0100, Carl Friedrich Bolz writes: >Gary Robinson wrote: >> 2) I understand Google is suppling $5000 per project ($4500 for the >> student). Is there any reason an outside company can't throw a little >> extra money into the pot for a particular task to sweeten the >> motivation for that task to be done in the SoC? Another thing to note -- it is possible to fund the PyPy project directly. If the idea of paying the people who are already working on the project -- but aren't students -- to do things you would like done has appeal, we can probably work something out. Laura Creighton

1 0

JIT SoC
by Gary Robinson March 28, 2008

March 28, 2008

I saw the suggestion "A good project would be to give the 386 backend a good refactoring, simplifying it, then add support for missing features like floating-point arithmetic." on the list of PyPy independent project suggestions. I'm wondering two things: 1) Is it realistically possible that after the SoC, if that work was taken on by someone, PyPy would be usable in an environment where high-speed floating-point calcs were needed? (The environment I have in mind wouldn't be putting much other stress on python -- it doesn't need a GUI, etc.) It sounds to me like it's far too big a project, but I don't know the existing code at all so I don't want to assume -- it would be great if it could be done. 2) I understand Google is suppling $5000 per project ($4500 for the student). Is there any reason an outside company can't throw a little extra money into the pot for a particular task to sweeten the motivation for that task to be done in the SoC? -- Gary Robinson CTO Emergent Music, LLC grobinson(a)goombah.com 207-942-3463 Company: http://www.goombah.com Blog: http://www.garyrobinson.net

2 1

enchancing flow objspace
by Maciek Fijalkowski March 28, 2008

March 28, 2008

what about making flow objspace understand xxx and XXX as global variables that always raise XxxError() instead of exploding? would simplify translation greatly. :.

3 2

summer of code 2008
by Jakub Gustak March 28, 2008

March 28, 2008

> I'm not sure about the technical details, but it sounds like a grea > idea. I think PyPy could benefit from more projects that explore > parallelism. Still. No sign of PyPy at the PSF ideas page. Is there any chance, PyPy will participate? __ Jakub Gustak

2 1

Re: [pypy-dev] Problems with modified socket.py
by Antonio Cuni March 23, 2008

March 23, 2008

Leandro Lameiro wrote: > Hi Antonio. > > My name is Leandro Lameiro and I've been a long time observer of PyPy. > This weekend I tried to see where I could contribute to PyPy, so I > decided to try and run some apps and see what is missing. Hi Leandro. Thank you for your interest in PyPy; every contribution is welcome, either patches or bug report! I'm CCing pypy-dev, so that other developers can read it as well. > The first thing I tried to run was ez_setup.py ( > http://peak.telecommunity.com/dist/ez_setup.py ), a simple script that > downloads and installs setuptools. It is a really short script but it > uses various modules such as md5, socket, urllib, os etc. > > After translating PyPy (with allworkingmodules and socket switches) I > found that ez_setup.py will raise and exception in PyPy but not in > CPython (tried with 2.4 and 2.5). > > Here is the traceback: > > (virtualenv)lameiro@liria:~/projetos/ez_setup$ python ez_setup.py > Downloading > http://pypi.python.org/packages/2.4/s/setuptools/setuptools-0.6c8-py2.4.egg > Traceback (most recent call last): > File "?", line 34, in run_toplevel > File "ez_setup.py", line 267, in <module> > main(sys.argv[1:]) > File "ez_setup.py", line 204, in main > os.unlink(egg) > File "ez_setup.py", line 198, in main > egg = download_setuptools(version, delay=0) > File "ez_setup.py", line 152, in download_setuptools > if src: src.close() > File "/home/lameiro/projetos/pypy-dist/lib-python/2.4.1/urllib.py", > line 804, in close > if self.fp: self.fp.close() > File > "/home/lameiro/projetos/pypy-dist/lib-python/modified-2.4.1/socket.py", > line 242, in close > s._drop() > AttributeError: 'HTTPResponse' object has no attribute '_drop' > (virtualenv)lameiro@liria:~/projetos/ez_setup$ > > After investigating a little, I found that you modified socket.py and > introduced this _drop call and also defined it in socket._closedsocket > class (an empty method), but that doesn't seem to be enough since other > types are going there (in this case HTTPResponse, but maybe others too). indeed, you are right; the relevant _drop method is defined in pypy-dist/pypy/module/_socket/interp_socket.py (it's called _drop_w), but I guess we should take care of cases like this in which the object doesn't provide a _drop method. > The revision in which this modification was introduced is > https://codespeak.net/viewvc/pypy/dist/lib-python/modified-2.4.1/socket.py?… > <https://codespeak.net/viewvc/pypy/dist/lib-python/modified-2.4.1/socket.py?…> > , and I suppose it was you because the commiter login is "ac". :) sorry, your guess is wrong :-). My login name is "antocuni", "ac" stands for Anders Chrigström. That's also why I don't know much about this and I'm CCing pypy-dev :-) > Anyway, thanks for the great work and I hope the next time I can > contribute a patch instead of a bug report. thank to you again for the help! Feel free to come to #pypy on freenode if you want to talk with us by IRC. ciao, Anto

1 0