Mailman 3 October 2005 - pypy-dev

even more import analysis
by Michael Hudson 06 Nov '05

06 Nov '05

So I've beaten at my importfun tool even more and now it can produce an html report about the import (lack of) structure of PyPy; for an example start here: http://starship.python.net/crew/mwh/importfunhtml/pypy.interpreter.argument… You might not want to look at the directory itself -- there's about 10000 files in it :) I also think there are some bugs & omissions so I'll try to work on this a little more tomorrow, but I've run out of time tonight. Enjoy! Cheers, mwh PS: generation took about 5 minutes, so keeping it up to date shouldn't be that hard. -- Haha! You had a *really* weak argument! <wink> -- Moshe Zadka, comp.lang.python

3 5

rel 0.8 coord meeting, 2pm GMT+1 monday 31 Oct 2005 on #pypy
by Samuele Pedroni 01 Nov '05

01 Nov '05

As already pointed out the scope of the 0.8 release is mainly: - Ship our integrated and now translated AST compiler - Ship the speed improvements we had since 0.7, which mainly means updating the documentation that discuss current speed - Ship a release in which the thunk space can also be translated The meeting is for people who think can spend some time helping with the release during the week. Agenda for the meeting (max ~1h): Status of the branch and regressions (max 15-20min) ------------------------------------------------- I have run tests and compliancy tests on the branch made on Friday: see issue 147 and http://www.strakt.com/~pedronis/testresult-pre0.8/ We have basically two regressions: - we have a failure in test_sort, because switching to using rpython lists directly has broken detecting changes during list sort in some situations. - translation with threads doesn't work at least because stack_too_big is not thread-safe, there may be other issues. Making stack_too_big thread-safe is not trivial x-platform wise. Discuss whether there are low-risk solutions that can be implemented in 1/2 days, otherwise I think the regressions are not show stoppers. Release tasks, mostly documentation updates (20-25min) ------------------------------------------------------ I have started adding issues to the tracker, under release 0.8, for updates/writing required for the release: - go through the list, see if something is amiss, possibly assign people to tasks and reviewing texts Packaging, uploading (15-20min) ---------------------------------- When and how to proceed making the candidate tarballs, uploading them and cutting the release. This is something so far Holger took care of, can this be delegated (or not for this release), how do we want to proceed? Thanks, Samuele.

2 1

release planning meeting on Monday
by Carl Friedrich Bolz 30 Oct '05

30 Oct '05

Hi PyPy-dev! On todays pypy-sync meeting (minutes at http://codespeak.net/pypy/extradoc/minute/pypy-sync-10-27-2005.txt) it was decided that we would try to make a release on friday next week. To plan the release work there will be a release planning meeting: Time: Monday 31th October 2005, 2pm-3pm (GMT + 2) Location: #pypy Samuele will moderate the meeting and send out the agenda on Sunday. The plan is that there should be a release branch tomorrow, Samuele will run the compliancy test over the weekend and we decide what needs to be done on Monday. Cheers, Carl Friedrich

3 2

Re: [Python-logic] choco or constraint programming in java
by Grégoire Dooms 28 Oct '05

28 Oct '05

cc-ing to pypy-dev as it is about a potential sprint. Nicolas Chauvat wrote: >During the last sprint we discussed sprinting in Belgium, which is >central for the partners, but since organising without anyone inviting >us was difficult, we postponed it. > >Do you think there would be interest at UCL for hosting a PyPy sprint >and getting a chance to meet with the Oz folks ? I would like that a >lot... > > I talked with Peter Van Roy and its researchers and I must say they are enthusiast about this idea. Hope it will work out. The best time would be between Jan 9th and Jan 30th (exams and 1week off for the students). I saw you will have a pypy project coordination meeting in Bruxelles in January so an idea would be to connect those two events. Best, -- Grégoire Dooms

4 3

Hi, I'm new here
by Paul deGrandis 28 Oct '05

28 Oct '05

All, Hi everyone, I'm Paul deGrandis, and I just joined the list. I plan on actively developing on the project and helping out anyway I can. Let me tell you a little about myself: I'm twenty years old and currently working my way through BS/MS dual degree in Software Engineering at Drexel University. I'm also a reasearcher and developer on the DARPA SAPIENT project. I have worked on numerous other open source project, but took a short break from FOSS software while I focused more on school and work. I'm ready to start working on another project now and am extremely excited about PyPy. If you would like to see an updated resume or anything, ask, I'll be happy to send one. I plan on working on some of the issues reported on the Issues section, but if anyone on the development team wants me to work on something specific, please email me. I can't wait to get working! Regards, Paul __________________________________ Yahoo! Mail - PC Magazine Editors' Choice 2005 http://mail.yahoo.com

4 5

import analysis of pypy
by Michael Hudson 28 Oct '05

28 Oct '05

If you've been reading pypy-svn you'll know I've written an import analysis tool (pypy/tool/importfun.py). While its implementation is currently completely inscrutable, here's some of its results. There are currently 242 import cycles in PyPy :(: http://starship.python.net/crew/mwh/importcycles.txt A 'summary' function gave this: the average module was imported 4.40 times imported 5.68 other modules defined 14.17 names there were 49 import *s the average one produced 5.33 names that were actually used Tests and tools aside, we have very few modules that aren't imported anywhere -- at least we're not too bad on this front :) And here's an analysis of the import *s in the code base (excluding tests, roughly speaking): http://starship.python.net/crew/mwh/importstar.txt Cheers, mwh -- <sjj> dash: you certainly are an enigma wrapped in a riddle wrapped in a hat. -- from Twisted.Quotes

2 1

python & llvm
by Simon Burton 28 Oct '05

28 Oct '05

Hi, I'd like to use llvm from python to generate dynamically: 1) llvm functions from python expressions, eg. an "apply" operation on a c-array of floats 2) callback functions 3) calls to an external dynamic loaded c-library I've looked at the llvm python wrappers but the size of the .so module is just too massive (120Mb). However, I don't think it is actually necessary for what I want to do, as I can just generate llvm code and send that to the llvm compiler (at runtime). This is what the pypy guys do. They mentioned maybe using the llvm-API, what for ? All that jazz is for people writing optimizing compilers, no ? Wouldn't it make more sense to do all that biz on the python level ? I did a preliminary test of calling a trivial c-function (in a dynamic lib) from llvm, by hardcoding the function address, and it seemed to work ok. So my main question is: am I on the right path; is this a reasonably cross-platform strategy ? Also i wanted to post to the pypy developers: there seems to be some mutual gravity here. Where I work we do heavy duty numeric processing with python (machine learning) [1]. Maybe I will get a chance to work on/with/steal from, pypy. (The guys here are absolutely obsessed with speed.) I just found out that the python AST branch has been merged, so i'll be investigating this aswell. bye for now, Simon. [1] https://lineal.developer.nicta.com.au/lineal -- Simon Burton, B.Sc. Licensed PO Box 8066 ANU Canberra 2601 Australia Ph. 61 02 6249 6940 http://arrowtheory.com

2 1

making a release branch for 0.8
by Samuele Pedroni 28 Oct '05

28 Oct '05

I'm about to make a release branch for 0.8 it will be at http://codespeak.net/svn/pypy/release/0.8.x Samuele.

1 0

Re: Question (please answer!)
by Richard Emslie 27 Oct '05

27 Oct '05

cc 'ing pypy-dev, since I cant get to #pypy these days. On Tue, 25 Oct 2005, Christian Tismer wrote: > Richard Emslie wrote: > >> Did see ll_osdefs.h breaks llvm in horrible ways... >> > > Aaahrrrgggg!!! > > Why that at all? > Just a brave part of the inclues, taken from CPython ( *gulp* ) > how can this hurt so much ? :-) Yes arrrggghhhh!! But no fault of yours - just that we use a (llvm) C compiler for externs. If a local compiler is not available, we go to a remote machine to generate llvm code. So local includes are not possible - hence in genllvm we manually include C files from #include statements. :-( Now listdir() uses opaque (dont know if you introduced this or someone else) - and llvm doesnt have proper support for opaque types... yet. So we cannot compile pypy right now. > >> will try to fix, but it is becoming more of a hassle trying to maintain >> llvm backend when it is not really giving much back. > Sorry for the negativeness. It would be interesting however to have the core developers opinion(s) about backends in general and whether it is believed maintaining it (the llvm one) is a worthwhile goal. > Sorry about that.I tried to make my live easier when I had to port > this listdir() mess, which does not fit winnows at all :-) > No need to apologize - the include fix was minor and opaque types are part of rpython and cannot be avoided. :-) Cheers & good nite, Richard

4 7

"this week in pypy"
by Michael Hudson 27 Oct '05

27 Oct '05

I've decided to write a summary each week (starting next week) of activity in pypy-land. I'd like people to help me by adding summary-worthy things to a file in svn; for more info see here: http://codespeak.net/pypy/dist/pypy/doc/weekly/ This isn't meant to discriminate against those without codepspeak logins -- if you want to note something either hassle someone on #pypy or send me an email. Thanks in advance! Cheers, mwh -- I really hope there's a catastrophic bug in some future e-mail program where if you try and send an attachment it cancels your ISP account, deletes your harddrive, and pisses in your coffee -- Adam Rixey

1 0