Mailman 3 September 2007 - pypy-dev

Re: [pypy-dev] pypy cleanup sprint
by Laura Creighton 13 Sep '07

13 Sep '07

In a message of Thu, 13 Sep 2007 15:21:18 +0200, Carl Friedrich Bolz writes: >Hi Paul, > >Paul deGrandis wrote: >> If at all possibly, can we also consider a place that has cheaper air >> fare deals? >> I'd really like to come to this sprint. > >The sprint venue (as discussed on IRC a while ago) is really quite >convenient for many, since Samuele, Laura, Jacob, Bea are living there, >many of us know the sprint venue and the city, etc. So I don't think >changing is a good idea. Maybe you can fly to Stockholm (or even >Kopenhagen) and take the train? That takes a few hours but is actually >very nice. > >Cheers, > >Carl Friedrich Planes may actually be cheaper. Where are you Paul? Laura

3 2

Re: [pypy-dev] pypy cleanup sprint
by Maciej Fijalkowski 13 Sep '07

13 Sep '07

Few points from me: * Kill half concrete wrapper (kill kill kill) * Kill opaque hacks on the C backend * decide how to implement constants in rffi * think about/remove orphaned parts * finish ll2ctypes, especially rffi_platform The last one is not very much cleanup, but is needed to finish transition from rctypes (like rsocket). Also, why not make sprint in Poland? (hint, hint, cheap place). PS. Greetings from china Cheers, fijal

3 4

cleanup sprint planning
by Carl Friedrich Bolz 11 Sep '07

11 Sep '07

Hi PyPy-Dev, there has been the vague plan for a cleanup sprint since the end of the EU project. It is becoming more concrete now: There is the vague idea of doing a sprint in October/November (probably November) at Laura's place in Gothenburg. To coordinate the possible times a bit, I have set up a doodle: http://doodle.ch/5awzskkt5mian94a Please enter your available times if you want to attend. Note that this sprint is going to be likely newbie-unfriendly because of its refactoring and cleanup focus. About the work for the sprint: this is a list of possible tasks (they may be too much and some may happen before the sprint), other suggestions are also possible as long as they involve "cleanups". Meta-Task --------- - Tracker-gardening - Start slowly thinking about 1.1 plans Translation toolchain --------------------- - finish rctypes removal - implement correct implicit GIL release - move away from implicit keepalives - raw_malloc should be split, rffi one should not use obmalloc (it's not thread-safe) - kill simulator, run the semi space tests on llinterp (may happen before) - have a moving gc correct version of the gc framework transformation - delegate finding type stuff like vtables etc to GC, cleaner interface for rtti, simplify translator/c/gc.py - think about approaches to id, especially concerning boehm, where the id will keep the object alive and concerning a moving GC - clean up the tangle of including headers in the C backend - review pdbplus, especially the graph commands, also in the light of https://codespeak.net/issue/pypy-dev/issue303 and the fact that we can have more than one translator/annotator around (with the timeshifter) - remember and fix the inlining exception bug Interpreter ----------- - there's a shadow tracking bug it seems - fix taint space - review the things implemented at applevel whether they are performance-critical JIT ------- - start writing real unit tests for the merging logic Please post all comments you might have about the planning or the tasks above. Looking forward to a productive sprint. Carl Friedrich & Samuele

1 0

NumPy in PyPy
by Greg Holling 11 Sep '07

11 Sep '07

Sorry if this has been posted before, but I didn't see anything in the archives. I see that there's been some recent activity in subversion related to numpy in pypy (specifically in pypy/dist/pypy/rpython/numpy). Can anybody on this list give me some more information about this development work (goals, status, etc.)? I have an upcoming project where it might be useful. Thanks! - Greg -

2 1

Testing generated JavaScript code
by James Ascroft-Leigh 09 Sep '07

09 Sep '07

Hi all, I am interested in the potential for PyPy to run python code within the web browser using, at least as a fallback, the JavaScript interpreter. I have my own project (very immature and not yet usable) to try an achieve just this goal and, like PyPy, my tests exercise the generated JavaScript code using an external interpreter (currently rhino). I am just writing to let you know about this great project I discovered the other day called jpype (http://jpype.sf.net/) that lets you interact with an in-process JVM. Now that the JVM includes the Rhino JS interpreter I am going to switch my tests over to using this to run my generated JS code. This allows a higher level interaction between the languages because you no longer have to serialise the output of the script to stdout and de-serialise it to compare to the expected output. Maybe PyPy could do the same? I might get time to post a patch for this but, as the attached snippet shows, it is not too hard to get started so somebody else might want to do it sooner. What do people think about adding a testing dependency on the JVM and this jpype package? Regards, James

1 0

Help raise Open Source visibility in the government area
by Godefroid Chapelle 06 Sep '07

06 Sep '07

Hi, ( If you have already received this email and have voted, thank you ! ) ( If you have not voted, and wanted to, this is a nice reminder. ) ( If you do not wish to vote, sorry for the noise. ) Please take 10 minutes to help raise open source profile in the government area by voting for the PloneGov project at the european egov awards. See below what PloneGov is and why you could vote to support open source even if you do not use Plone. Anyone from any country can vote, you don't have to live in the EU. And please vote *now*, the deadline is almost upon us **SEPTEMBER 7** — you are likely to forget if you don't set aside 10 minutes now. This process is unfortunately a bit involved — since we're dealing with governments — but shouldn't take longer than 10 minutes to complete. Step-by-step: 1. Go to http://www.epractice.eu/register 2. Wait for the email confirmation and log in. 3. Fill out your profile: http://www.epractice.eu/myprofile/personal http://www.epractice.eu/myprofile/professional http://www.epractice.eu/myprofile/contact (If you don't want to provide professional info, we suggest you add "I just want to vote" in the required fields) 4. Vote for PloneGov by clicking here: http://www.epractice.eu/do/voteinspiring/1026 Thanks! More information below, if you're interested in what PloneGov is. _____________________________________ What is PloneGov? PloneGov aims to create a collaborative software ecosystem where governmental organizations work together, in close relationship with SMEs, to share the cost of software development and enhanced capabilities. This new initiative started by small towns now reaches all levels of public administration and spans over 13 countries on 3 continents. http://www.plonegov.org Your vote for the "Most innovative good practice" award is a unique opportunity to reach politicians and decision makers. Winning this award will highlight the maturity achieved by the Open Source collaborative model as a pragmatic way to tackle IT challenges in society. You may vote until Friday, September 7th. The Public Sector has an important influence on many sectors of society. A wider Open Source acceptance in this field is strategic to all Open Source users and advocates. Users in sectors such as Education, NGOs and businesses would directly benefit from broader adoption. It would result in a growing pool of available software, tools, and developers. Beyond the practical example of PloneGov, your vote contributes to support Open Source values that are based on collaboration, openness and fair trade. It is an opportunity to support Open Source as an innovative model of IT development in which the Public Sector, NGOs and Education can work together and all of society will benefit. -- Godefroid Chapelle (aka __gotcha)- BubbleNet http://bubblenet.be

1 0

Branch Merged!
by Armin Rigo 05 Sep '07

05 Sep '07

Hi all, We've just merged the pypy-more-rtti-inprogress branch. For those of you still working in there - I'm thining about Simon in particular - you need to switch back to the regular trunk. In principle, you can issue the following command in your working copy even if there are local changes, which should stay untouched as long as they don't conflict (but make a backup, just in case!): In a checkout of pypy/dist/pypy, in the 'pypy' directory: svn switch http://codespeak.net/svn/pypy/dist/pypy This branch was a port of all low-level functions of the os module to use the new RFFI interface. The branch also contains progress made on RFFI itself. I've tested and fixed the branch yesterday for Windows XP, so it should compile there as well. A bientot, Armin.

1 0