From lac at openend.se Tue Jun 2 19:33:23 2009 From: lac at openend.se (Laura Creighton) Date: Tue, 2 Jun 2009 19:33:23 +0200 Subject: [pypy-dev] crosstwine linker Message-ID: <200906021733.n52HXNsZ008975@theraft.openend.se> Anybody heard of this? http://crosstwine.com/linker/index.html Damien Diederen is giving a talk about speeding up python at EuroSciPy, and since this is his company, it should be about this. Laura From anto.cuni at gmail.com Tue Jun 2 20:46:04 2009 From: anto.cuni at gmail.com (Antonio Cuni) Date: Tue, 02 Jun 2009 20:46:04 +0200 Subject: [pypy-dev] crosstwine linker In-Reply-To: <200906021733.n52HXNsZ008975@theraft.openend.se> References: <200906021733.n52HXNsZ008975@theraft.openend.se> Message-ID: <4A25736C.7070109@gmail.com> Laura Creighton wrote: > Anybody heard of this? > http://crosstwine.com/linker/index.html > > Damien Diederen is giving a talk about speeding up python at EuroSciPy, > and since this is his company, it should be about this. I skimmed over the website and read the whitepaper few weeks ago. Honestly, I don't understand if and how they manage to speedup python. In fact, they show results for only 4 benchmarks: one is a version of bpnn (which we also have, somewhere) modified to be "static enough" to be compiled by shedskin; the other three are really microbenchmarks. Since he doesn't show any other benchmark, I wonder if his product gives any speedup at all on real programs. In other words: I think we should also point out the few benchmarks where pypy is already faster and start making money on it :-) From fijall at gmail.com Tue Jun 2 20:56:55 2009 From: fijall at gmail.com (Maciej Fijalkowski) Date: Tue, 2 Jun 2009 12:56:55 -0600 Subject: [pypy-dev] crosstwine linker In-Reply-To: <4A25736C.7070109@gmail.com> References: <200906021733.n52HXNsZ008975@theraft.openend.se> <4A25736C.7070109@gmail.com> Message-ID: <693bc9ab0906021156o7627b0c2p8feff1c8fa52bf88@mail.gmail.com> As usual with such products, we don't know if they implemented 100% python or 99%. Can they run existing apps? Basically, the whitepaper as it is, is not a scientific experiment because one is unable to reproduce results. I would simply ignore it until someone has something to show. Cheers, fijal On Tue, Jun 2, 2009 at 12:46 PM, Antonio Cuni wrote: > Laura Creighton wrote: >> Anybody heard of this? >> http://crosstwine.com/linker/index.html >> >> Damien Diederen is giving a talk about speeding up python at EuroSciPy, >> and since this is his company, it should be about this. > > I skimmed over the website and read the whitepaper few weeks ago. ?Honestly, I > don't understand if and how they manage to speedup python. ?In fact, they show > results for only 4 benchmarks: one is a version of bpnn (which we also have, > somewhere) modified to be "static enough" to be compiled by shedskin; the > other three are really microbenchmarks. > > Since he doesn't show any other benchmark, I wonder if his product gives any > speedup at all on real programs. > > > In other words: I think we should also point out the few benchmarks where pypy > is already faster and start making money on it :-) > _______________________________________________ > pypy-dev at codespeak.net > http://codespeak.net/mailman/listinfo/pypy-dev > From cfbolz at gmx.de Tue Jun 2 22:14:59 2009 From: cfbolz at gmx.de (Carl Friedrich Bolz) Date: Tue, 02 Jun 2009 22:14:59 +0200 Subject: [pypy-dev] crosstwine linker In-Reply-To: <693bc9ab0906021156o7627b0c2p8feff1c8fa52bf88@mail.gmail.com> References: <200906021733.n52HXNsZ008975@theraft.openend.se> <4A25736C.7070109@gmail.com> <693bc9ab0906021156o7627b0c2p8feff1c8fa52bf88@mail.gmail.com> Message-ID: <4A258843.2010901@gmx.de> Maciej Fijalkowski wrote: > As usual with such products, we don't know if they implemented 100% > python or 99%. > Can they run existing apps? > > Basically, the whitepaper as it is, is not a scientific experiment > because one is > unable to reproduce results. I would simply ignore it until someone > has something > to show. a) they provide a binary b) they say that they can run the CPython test-suite Let's please not randomly bash projects that we don't know enough about. The whitepaper seems to make sense, technically. It would be a lot of work to implement, but maybe somebody sat down and did just that. Cheers, Carl Friedrich From fijall at gmail.com Tue Jun 2 22:26:01 2009 From: fijall at gmail.com (Maciej Fijalkowski) Date: Tue, 2 Jun 2009 14:26:01 -0600 Subject: [pypy-dev] crosstwine linker In-Reply-To: <4A258843.2010901@gmx.de> References: <200906021733.n52HXNsZ008975@theraft.openend.se> <4A25736C.7070109@gmail.com> <693bc9ab0906021156o7627b0c2p8feff1c8fa52bf88@mail.gmail.com> <4A258843.2010901@gmx.de> Message-ID: <693bc9ab0906021326x4c127d45w38e89f8b3baadc3c@mail.gmail.com> On Tue, Jun 2, 2009 at 2:14 PM, Carl Friedrich Bolz wrote: > Maciej Fijalkowski wrote: >> >> As usual with such products, we don't know if they implemented 100% >> python or 99%. >> Can they run existing apps? >> >> Basically, the whitepaper as it is, is not a scientific experiment >> because one is >> unable to reproduce results. I would simply ignore it until someone >> has something >> to show. > > a) they provide a binary Link? > b) they say that they can run the CPython test-suite > > Let's please not randomly bash projects that we don't know enough about. The > whitepaper seems to make sense, technically. It would be a lot of work to > implement, but maybe somebody sat down and did just that. > > Cheers, > > Carl Friedrich > From fijall at gmail.com Tue Jun 2 22:27:14 2009 From: fijall at gmail.com (Maciej Fijalkowski) Date: Tue, 2 Jun 2009 14:27:14 -0600 Subject: [pypy-dev] crosstwine linker In-Reply-To: <4A258843.2010901@gmx.de> References: <200906021733.n52HXNsZ008975@theraft.openend.se> <4A25736C.7070109@gmail.com> <693bc9ab0906021156o7627b0c2p8feff1c8fa52bf88@mail.gmail.com> <4A258843.2010901@gmx.de> Message-ID: <693bc9ab0906021327u47a892ddu5df63d82bda0af12@mail.gmail.com> On Tue, Jun 2, 2009 at 2:14 PM, Carl Friedrich Bolz wrote: > Maciej Fijalkowski wrote: >> >> As usual with such products, we don't know if they implemented 100% >> python or 99%. >> Can they run existing apps? >> >> Basically, the whitepaper as it is, is not a scientific experiment >> because one is >> unable to reproduce results. I would simply ignore it until someone >> has something >> to show. > > a) they provide a binary oops, found. > b) they say that they can run the CPython test-suite > > Let's please not randomly bash projects that we don't know enough about. The > whitepaper seems to make sense, technically. It would be a lot of work to > implement, but maybe somebody sat down and did just that. > > Cheers, > > Carl Friedrich > From fijall at gmail.com Tue Jun 2 22:53:10 2009 From: fijall at gmail.com (Maciej Fijalkowski) Date: Tue, 2 Jun 2009 14:53:10 -0600 Subject: [pypy-dev] crosstwine linker In-Reply-To: <693bc9ab0906021327u47a892ddu5df63d82bda0af12@mail.gmail.com> References: <200906021733.n52HXNsZ008975@theraft.openend.se> <4A25736C.7070109@gmail.com> <693bc9ab0906021156o7627b0c2p8feff1c8fa52bf88@mail.gmail.com> <4A258843.2010901@gmx.de> <693bc9ab0906021327u47a892ddu5df63d82bda0af12@mail.gmail.com> Message-ID: <693bc9ab0906021353p1add08cagee67c663e02afb8b@mail.gmail.com> Ok, so I did my homework and performed a couple of checks: * it seems to mostly work on a couple of examples that I checked * I was a bit comparing aples to oranges (python 2.6 vs xtpython 3.1), but it basically gives a bit of speedup (10%-2x) over loops which do stuff. * checking it on almost anything but stuff written by hand makes no sense since it's 3.1 * result binary is 64bit only. On Tue, Jun 2, 2009 at 2:27 PM, Maciej Fijalkowski wrote: > On Tue, Jun 2, 2009 at 2:14 PM, Carl Friedrich Bolz wrote: >> Maciej Fijalkowski wrote: >>> >>> As usual with such products, we don't know if they implemented 100% >>> python or 99%. >>> Can they run existing apps? >>> >>> Basically, the whitepaper as it is, is not a scientific experiment >>> because one is >>> unable to reproduce results. I would simply ignore it until someone >>> has something >>> to show. >> >> a) they provide a binary > > oops, found. > >> b) they say that they can run the CPython test-suite >> >> Let's please not randomly bash projects that we don't know enough about. The >> whitepaper seems to make sense, technically. It would be a lot of work to >> implement, but maybe somebody sat down and did just that. >> >> Cheers, >> >> Carl Friedrich >> > From bbuck3898 at emailias.com Wed Jun 3 19:16:15 2009 From: bbuck3898 at emailias.com (Bryan Buck) Date: Wed, 03 Jun 2009 13:16:15 -0400 Subject: [pypy-dev] Calling PyPy interpreter from C? Message-ID: <1244049375.13273.1318611689@webmail.messagingengine.com> Hi, I have a question about PyPy - I hope this mailing list is the right place for it, let me know if not. Is there any way to link PyPy's Python interpreter with a C program and call it from C? I'm interested in embedding the interpreter in another program as its scripting engine. (The reason I'm looking at PyPy instead of CPython is that I'd like to use the sandboxing feature.) Thanks for any information you can give me. - Bryan Buck From benjamin at python.org Wed Jun 3 19:17:58 2009 From: benjamin at python.org (Benjamin Peterson) Date: Wed, 3 Jun 2009 12:17:58 -0500 Subject: [pypy-dev] Calling PyPy interpreter from C? In-Reply-To: <1244049375.13273.1318611689@webmail.messagingengine.com> References: <1244049375.13273.1318611689@webmail.messagingengine.com> Message-ID: <1afaf6160906031017xcaf7260qc3eacb0bd0ba5bb8@mail.gmail.com> 2009/6/3 Bryan Buck : > Hi, > > I have a question about PyPy - I hope this mailing list is the right > place for it, let me know if not. > > Is there any way to link PyPy's Python interpreter with a C program and > call it from C? I'm interested in embedding the interpreter in another > program as its scripting engine. (The reason I'm looking at PyPy instead > of CPython is that I'd like to use the sandboxing feature.) Not easily. The best way would probably be to run pypy in a subprocess. > > Thanks for any information you can give me. -- Regards, Benjamin From santagada at gmail.com Wed Jun 3 19:49:09 2009 From: santagada at gmail.com (Leonardo Santagada) Date: Wed, 3 Jun 2009 14:49:09 -0300 Subject: [pypy-dev] Calling PyPy interpreter from C? In-Reply-To: <1244049375.13273.1318611689@webmail.messagingengine.com> References: <1244049375.13273.1318611689@webmail.messagingengine.com> Message-ID: <53F1C367-F46F-4E20-BD57-2D8DA80AB718@gmail.com> On Jun 3, 2009, at 2:16 PM, Bryan Buck wrote: > Hi, > > I have a question about PyPy - I hope this mailing list is the right > place for it, let me know if not. > > Is there any way to link PyPy's Python interpreter with a C program > and > call it from C? I'm interested in embedding the interpreter in another > program as its scripting engine. (The reason I'm looking at PyPy > instead > of CPython is that I'd like to use the sandboxing feature.) I think that specially because of the sandbox you don't want to link the interpreter with your c program. The whole idea of the sandbox is to comunicate with the intepreter only thru a controlling program that uses stdin/stdout to control the pypy interpreter in a subprocess. Take a look at the sandboxing page: http://codespeak.net/pypy/dist/pypy/doc/sandbox.html see: ''we can translate PyPy to a special "pypy-c-sandbox" executable, which is safe in the sense that it doesn't do any library or system calls - instead, whenever it would like to perform such an operation, it marshals the operation name and the arguments to its stdout and it waits for the marshalled result on its stdin. This pypy-c-sandbox process is meant to be run by an outer "controller" program that answers these operation requests.'' -- Leonardo Santagada santagada at gmail.com From bbuck3898 at emailias.com Wed Jun 3 20:12:03 2009 From: bbuck3898 at emailias.com (Bryan Buck) Date: Wed, 03 Jun 2009 14:12:03 -0400 Subject: [pypy-dev] Calling PyPy interpreter from C? In-Reply-To: <1afaf6160906031017xcaf7260qc3eacb0bd0ba5bb8@mail.gmail.com> References: <1244049375.13273.1318611689@webmail.messagingengine.com> <1afaf6160906031017xcaf7260qc3eacb0bd0ba5bb8@mail.gmail.com> Message-ID: <1244052723.25081.1318619587@webmail.messagingengine.com> Thank you for your reply. A subprocess is a possibility, although I'd like to avoid it if I can. Of course I was hoping there was some already- available way to do this that I had overlooked, but if necessary modifying PyPy isn't totally out of the question. How difficult is not easily? I guess I would have to change the translator to produce a library instead of an executable, and create an API for calling the interpreter from C? It does seem like a somewhat involved project, but I'm just trying to gauge the relative difficulty vs. the other possibilities I have in mind (all of which would also take significant effort). Thanks, Bryan Buck On Wed, 03 Jun 2009 12:17 -0500, "Benjamin Peterson" wrote: > 2009/6/3 Bryan Buck : > > Hi, > > > > I have a question about PyPy - I hope this mailing list is the right > > place for it, let me know if not. > > > > Is there any way to link PyPy's Python interpreter with a C program > > and call it from C? I'm interested in embedding the interpreter in > > another program as its scripting engine. (The reason I'm looking at > > PyPy instead of CPython is that I'd like to use the sandboxing > > feature.) > > Not easily. The best way would probably be to run pypy in a > subprocess. > > > > > Thanks for any information you can give me. > > -- > Regards, Benjamin th From benjamin at python.org Wed Jun 3 20:34:35 2009 From: benjamin at python.org (Benjamin Peterson) Date: Wed, 3 Jun 2009 13:34:35 -0500 Subject: [pypy-dev] Calling PyPy interpreter from C? In-Reply-To: <1244052723.25081.1318619587@webmail.messagingengine.com> References: <1244049375.13273.1318611689@webmail.messagingengine.com> <1afaf6160906031017xcaf7260qc3eacb0bd0ba5bb8@mail.gmail.com> <1244052723.25081.1318619587@webmail.messagingengine.com> Message-ID: <1afaf6160906031134u525f5153gaf31a9a3a5985411@mail.gmail.com> 2009/6/3 Bryan Buck : > Thank you for your reply. A subprocess is a possibility, although I'd > like to avoid it if I can. Of course I was hoping there was some already- > available way to do this that I had overlooked, but if necessary > modifying PyPy isn't totally out of the question. How difficult is > not easily? I guess I would have to change the translator to produce > a library instead of an executable, and create an API for calling the > interpreter from C? It does seem like a somewhat involved project, > but I'm just trying to gauge the relative difficulty vs. the other > possibilities I have in mind (all of which would also take > significant effort). Because the translation framework bridges such a large gap in languages, it would take significant effort to pin down a human-usable PyPy API at the C level. I don't know what your other options are, but they are almost certainly easier than creating a C-API for PyPy. :) -- Regards, Benjamin From bbuck3898 at emailias.com Wed Jun 3 20:46:50 2009 From: bbuck3898 at emailias.com (Bryan Buck) Date: Wed, 03 Jun 2009 14:46:50 -0400 Subject: [pypy-dev] Calling PyPy interpreter from C? In-Reply-To: <1afaf6160906031134u525f5153gaf31a9a3a5985411@mail.gmail.com> References: <1244049375.13273.1318611689@webmail.messagingengine.com><1afaf6160906031017xcaf7260qc3eacb0bd0ba5bb8@mail.gmail.com><1244052723.25081.1318619587@webmail.messagingengine.com> <1afaf6160906031134u525f5153gaf31a9a3a5985411@mail.gmail.com> Message-ID: <1244054810.32105.1318629511@webmail.messagingengine.com> Okay, thanks. I guess I'll look into either running the PyPy interpreter as a subprocess or using something else. Thanks again! - Bryan Buck On Wed, 03 Jun 2009 13:34 -0500, "Benjamin Peterson" wrote: > 2009/6/3 Bryan Buck : > > Thank you for your reply. A subprocess is a possibility, although > > I'd like to avoid it if I can. Of course I was hoping there was some > > already- available way to do this that I had overlooked, but if > > necessary modifying PyPy isn't totally out of the question. How > > difficult is not easily? I guess I would have to change the > > translator to produce a library instead of an executable, and create > > an API for calling the interpreter from C? It does seem like a > > somewhat involved project, but I'm just trying to gauge the relative > > difficulty vs. the other possibilities I have in mind (all of which > > would also take significant effort). > > Because the translation framework bridges such a large gap in > languages, it would take significant effort to pin down a human-usable > PyPy API at the C level. I don't know what your other options are, but > they are almost certainly easier than creating a C-API for PyPy. :) > > > -- > Regards, Benjamin From intelliyole at gmail.com Sat Jun 6 10:19:32 2009 From: intelliyole at gmail.com (Dmitry Jemerov) Date: Sat, 6 Jun 2009 12:19:32 +0400 Subject: [pypy-dev] Generating pypy-cli and pypy-jvm fails under Win32 Message-ID: <60423dd30906060119q6f1e0ab0od4644df3ae1af09@mail.gmail.com> Hello everyone, I'm trying to get my head around PyPy (hoping maybe to participate in the PyPy sprint at EuroPython), and following the getting-started guide on the Web site. Translating pypy with the C backend worked with no problems on my machine, but neither the JVM nor CLI backends work. (I tried both the pypy-dist version and the SVN trunk.) The stacktrace for the error is: Traceback (most recent call last): File "translate.py", line 273, in main drv.proceed(goals) File "C:\Src\pypy\pypy\translator\driver.py", line 704, in proceed return self._execute(goals, task_skip = self._maybe_skip()) File "C:\Src\pypy\pypy\translator\tool\taskengine.py", line 116, in _execute res = self._do(goal, taskcallable, *args, **kwds) File "C:\Src\pypy\pypy\translator\driver.py", line 267, in _do res = func() File "C:\Src\pypy\pypy\translator\driver.py", line 395, in task_backendopt_ootype backend_optimizations(self.translator) File "C:\Src\pypy\pypy\translator\backendopt\all.py", line 80, in backend_optimizations inline_heuristic=heuristic) File "C:\Src\pypy\pypy\translator\backendopt\all.py", line 156, in inline_malloc_removal_phase call_count_pred=call_count_pred) File "C:\Src\pypy\pypy\translator\backendopt\inline.py", line 779, in auto_inline_graphs call_count_pred=call_count_pred) File "C:\Src\pypy\pypy\translator\backendopt\inline.py", line 750, in auto_inlining call_count_pred, cleanup=False) File "C:\Src\pypy\pypy\translator\backendopt\inline.py", line 108, in inline_function return inliner.inline_all() File "C:\Src\pypy\pypy\translator\backendopt\inline.py", line 217, in inline_all self.inline_once(block, index_operation) File "C:\Src\pypy\pypy\translator\backendopt\inline.py", line 244, in inline_once self.translator, self.raise_analyzer): File "C:\Src\pypy\pypy\translator\backendopt\inline.py", line 160, in any_call_to_raising_graphs if does_raise_directly(graph_or_something, raise_analyzer): File "C:\Src\pypy\pypy\translator\backendopt\inline.py", line 153, in does_raise_directly if raise_analyzer.can_raise(op): File "C:\Src\pypy\pypy\translator\backendopt\canraise.py", line 33, in can_raise return self.analyze(op, seen) File "C:\Src\pypy\pypy\translator\backendopt\graphanalyze.py", line 38, in analyze return self.analyze_external_call(op) File "C:\Src\pypy\pypy\translator\backendopt\canraise.py", line 21, in analyze_external_call fnobj = deref(op.args[0].value) File "C:\Src\pypy\pypy\rpython\typesystem.py", line 153, in deref assert isinstance(ootype.typeOf(obj), ootype.OOType) AssertionError When I evaluate 'obj' with the pdb, I get: (Pdb+) p obj <* fn CryptGenRandom> As far as I understand, the problem happens because the 'posix' module under Win32 implements 'urandom' by calling the native CryptGetRandom function through rffi, but there is no mapping for this call under the JVM and CLI backends. What would be the proper way to fix this problem? (Or maybe I'm doing something that's not supposed to work at all at this stage of the project?) (Looks like my question should be at least partly answered by the "OO backends" section of the document at http://codespeak.net/pypy/dist/pypy/doc/rffi.html but unfortunately all it says is "XXX to be written".) Thanks in advance for your help! Dmitry -------------- next part -------------- An HTML attachment was scrubbed... URL: From dd at crosstwine.com Sat Jun 6 14:48:12 2009 From: dd at crosstwine.com (Damien Diederen) Date: Sat, 06 Jun 2009 14:48:12 +0200 Subject: [pypy-dev] crosstwine linker In-Reply-To: <693bc9ab0906021353p1add08cagee67c663e02afb8b@mail.gmail.com> (Maciej Fijalkowski's message of "Tue, 2 Jun 2009 14:53:10 -0600") References: <200906021733.n52HXNsZ008975@theraft.openend.se> <4A25736C.7070109@gmail.com> <693bc9ab0906021156o7627b0c2p8feff1c8fa52bf88@mail.gmail.com> <4A258843.2010901@gmx.de> <693bc9ab0906021327u47a892ddu5df63d82bda0af12@mail.gmail.com> <693bc9ab0906021353p1add08cagee67c663e02afb8b@mail.gmail.com> Message-ID: <87d49hfrzn.fsf@keem.bcc> Hi, I'm the author of the CrossTwine Linker framework. I hope most of you will be at EuroSciPy so that we get to meet in-person; in the meantime, I'm going to try to cover some of the questions raised in this thread: Antonio Cuni writes: > Laura Creighton wrote: >> Anybody heard of this? >> http://crosstwine.com/linker/index.html >> >> Damien Diederen is giving a talk about speeding up python at EuroSciPy, >> and since this is his company, it should be about this. It is. I'm going to talk about some of the ideas, inner workings, and methodologies underlying the project. I will also discuss some of the challenges of the current Python environment, and how we overcome them?the ones we have figured out, that is. [ Feel free to suggest topics for me to cover if you are interested in a particular aspect! ] > I skimmed over the website and read the whitepaper few weeks ago. > Honestly, I don't understand if and how they manage to speedup python. > In fact, they show results for only 4 benchmarks: one is a version of > bpnn (which we also have, somewhere) modified to be "static enough" to > be compiled by shedskin; the other three are really microbenchmarks. You are perfectly right: alpha 1 focused on microbenchmarks (one has to start somewhere, right? :) Cutting new releases, updating the website and publishing a not-so-lame set of benchmarks are very high on my TODO list?but unfortunately not quite at the top yet. > Since he doesn't show any other benchmark, I wonder if his product > gives any speedup at all on real programs. Real programs are very much on the radar screen now that the framework is getting more mature. We are seeing interesting progress on the development versions, even if there is still a *lot* of work to do. On the positive side, the growing set of components at our disposal seems to ?easily? accommodate more and more specialization scenarios. >>> Maciej Fijalkowski wrote: >>>> As usual with such products, we don't know if they implemented 100% >>>> python or 99%. Can they run existing apps? The Python demo is a derivative of the default interpreter, and we were very careful to avoid breakage while integrating pieces of the ?bolt-on? JIT and utilities. We are actually taking quite a speed hit to accommodate some of the corner cases, and hope to suppress some of that overhead by being a bit ?smarter? in the future. So we should be able to run 100% of existing apps, and differences in behaviour are to be considered bugs. Maciej Fijalkowski writes: > Ok, so I did my homework and performed a couple of checks: > > * it seems to mostly work on a couple of examples that I checked > > * I was a bit comparing aples to oranges (python 2.6 vs xtpython 3.1), > but it basically gives a bit of speedup (10%-2x) over loops which do stuff. We have a 2.6 version in the labs, to be included in the next (alpha 2) release. This will make comparisions more interesting, and allow us to focus more on real-world scenarios. > * checking it on almost anything but stuff written by hand makes no sense > since it's 3.1 Tell me about it! :) [ Starting with 3.x was an ?accident.? I am a relative newcomer to the Python world, and it looked like everybody was about to jump to 3.0 when we got started. Oh well? ] > * result binary is 64bit only. Yes, and while we could definitely develop a 32 bit version or other architecture backends, these are not currently part of the roadmap. We're firmly in x86-64 land for the time being, except if somebody steers us away with a particular use case. Cheers, Damien -- http://crosstwine.com "Strong Opinions, Weakly Held" -- Bob Johansen From fijall at gmail.com Mon Jun 8 05:23:34 2009 From: fijall at gmail.com (Maciej Fijalkowski) Date: Sun, 7 Jun 2009 21:23:34 -0600 Subject: [pypy-dev] [pypy-svn] r65652 - pypy/branch/pyjitpl5-experiments/pypy/jit/metainterp In-Reply-To: <20090607221305.CC34A16843D@codespeak.net> References: <20090607221305.CC34A16843D@codespeak.net> Message-ID: <693bc9ab0906072023t731b8f9aq5c37deef77fe5a24@mail.gmail.com> On Sun, Jun 7, 2009 at 4:13 PM, wrote: > Author: antocuni > Date: Mon Jun ?8 00:13:03 2009 > New Revision: 65652 > > Modified: > ? pypy/branch/pyjitpl5-experiments/pypy/jit/metainterp/optimize3.py > Log: > more refactoring&simplification: don't track the constness of nodes, as nobody > is using it so far. ?The idea is that optimizatons relying on the constness > should be done during the optimize_operation phase, not during the find_node > phase. ?I'm not 100% sure this will cover all possible case though, so maybe > we will need to reintroduce this later. > That is a bit of a problem, since a lot of things can only happen if constant folding was already done. A good example is a guard_value followed by getarrayitem_gc. getarrayitem_gc needs to have constant index for anything like virtuals or virtualizables. If it's a non-constant index, nodes will be incorrect (you cannot delay putting stuff in it's dictionary). From fijall at gmail.com Mon Jun 8 05:29:36 2009 From: fijall at gmail.com (Maciej Fijalkowski) Date: Sun, 7 Jun 2009 21:29:36 -0600 Subject: [pypy-dev] pypy on language shootout. Message-ID: <693bc9ab0906072029j3f5f0e16h5fdcd1eec43da980@mail.gmail.com> Hey. I have a quick question. Does any of you have account and can post stuff on shootout.alioth.debian.org? I would like to post a comment that I completely cannot repeat the measurments of pypy vs cpython speed on any computer that I have access to. Cheers, fijal From fijall at gmail.com Mon Jun 8 05:32:47 2009 From: fijall at gmail.com (Maciej Fijalkowski) Date: Sun, 7 Jun 2009 21:32:47 -0600 Subject: [pypy-dev] pypy on language shootout. In-Reply-To: <693bc9ab0906072029j3f5f0e16h5fdcd1eec43da980@mail.gmail.com> References: <693bc9ab0906072029j3f5f0e16h5fdcd1eec43da980@mail.gmail.com> Message-ID: <693bc9ab0906072032n311e9ee1hcc079b7f02de9895@mail.gmail.com> Grrr, sent too fast. I tried to create an account there to complain, but completely failed to do so. On Sun, Jun 7, 2009 at 9:29 PM, Maciej Fijalkowski wrote: > Hey. > > I have a quick question. Does any of you have account and can post > stuff on shootout.alioth.debian.org? > > I would like to post a comment that I completely cannot repeat the > measurments of pypy vs cpython > speed on any computer that I have access to. > > Cheers, > fijal > From anto.cuni at gmail.com Mon Jun 8 10:17:52 2009 From: anto.cuni at gmail.com (Antonio Cuni) Date: Mon, 08 Jun 2009 10:17:52 +0200 Subject: [pypy-dev] [pypy-svn] r65652 - pypy/branch/pyjitpl5-experiments/pypy/jit/metainterp In-Reply-To: <693bc9ab0906072023t731b8f9aq5c37deef77fe5a24@mail.gmail.com> References: <20090607221305.CC34A16843D@codespeak.net> <693bc9ab0906072023t731b8f9aq5c37deef77fe5a24@mail.gmail.com> Message-ID: <4A2CC930.80403@gmail.com> Maciej Fijalkowski wrote: >> more refactoring&simplification: don't track the constness of nodes, as nobody >> is using it so far. The idea is that optimizatons relying on the constness >> should be done during the optimize_operation phase, not during the find_node >> phase. I'm not 100% sure this will cover all possible case though, so maybe >> we will need to reintroduce this later. >> > > That is a bit of a problem, since a lot of things can only happen if > constant folding was already > done. A good example is a guard_value followed by getarrayitem_gc. > getarrayitem_gc needs > to have constant index for anything like virtuals or virtualizables. > If it's a non-constant index, > nodes will be incorrect (you cannot delay putting stuff in it's dictionary). uhm, now that I read that port of optimize.py more carefully, I think you are right. It's a bit unfortunate that we need two almost identical features in two different places, though :-/ From lac at openend.se Tue Jun 9 20:04:56 2009 From: lac at openend.se (Laura Creighton) Date: Tue, 9 Jun 2009 20:04:56 +0200 Subject: [pypy-dev] news flash -- Europython to have a Python VM Panel talk Message-ID: <200906091804.n59I4uUp029962@theraft.openend.se> Who wants to be on it? This question is not limited to PyPyers -- I suspect that Frank Wierzbicki would like to be on this. Laura From cfbolz at gmx.de Wed Jun 10 14:01:25 2009 From: cfbolz at gmx.de (Carl Friedrich Bolz) Date: Wed, 10 Jun 2009 14:01:25 +0200 Subject: [pypy-dev] news flash -- Europython to have a Python VM Panel talk In-Reply-To: <200906091804.n59I4uUp029962@theraft.openend.se> References: <200906091804.n59I4uUp029962@theraft.openend.se> Message-ID: <4A2FA095.7070704@gmx.de> Laura Creighton wrote: > Who wants to be on it? This question is not limited to PyPyers -- > I suspect that Frank Wierzbicki would like to be on this. I would like to be on the panel. Cheers, Carl Friedrich From lac at openend.se Fri Jun 12 08:08:24 2009 From: lac at openend.se (Laura Creighton) Date: Fri, 12 Jun 2009 08:08:24 +0200 Subject: [pypy-dev] From europython Message-ID: <200906120608.n5C68OXX026892@theraft.openend.se> ------- Forwarded Message Return-Path: europython-contact-bounces+lac=openend.se at python.org Delivery-Date: Fri Jun 12 01:57:27 2009 From: John Pinner To: Graham Manwell Cc: europython-talks at python.org, europython-contact at python.org Subject: Re: [europython-contact] EuroPython2009 Volunteering Hello Graham, 2009/6/12 Graham Manwell : > Hi John, > Sorry not to do this via the wiki - can't seem to get that to work > from home at the moment. Strange. >??I'm at EuroPython09 from 27/06/09 and am > happy to try to chair any sessions or tutorials you might have free > slots for - my field is compilers so if I chaired stuff in that area > at least I'd.have some chance of asking intelligent questions - > although PyPy makes my brain hurt. ??Also happy to offer any help with > bag stuffing, registration and so on. ??Let me know what you need. Thanks, that's great. Best wishes, John - -- > Cheers, > Graham > > J. Graham Manwell > Lecturer in Computer Science, > University of the West of Scotland. > > t: 0141 848 3545/0141 632 2928 > e: jgmanwell at gmail.com/manw-ci0 at uws.ac.uk ------- End of Forwarded Message I wonder if 'making my brain hurt' is to be considered positive in this context? Laura From cfbolz at gmx.de Sun Jun 21 17:07:13 2009 From: cfbolz at gmx.de (Carl Friedrich Bolz) Date: Sun, 21 Jun 2009 17:07:13 +0200 Subject: [pypy-dev] Europython slides Message-ID: <4A3E4CA1.7060207@gmx.de> Hi all, just wanted to remind that we are asked to submit the Europython slides in advance, I think the date was March 22nd (?). Unfortunately I am quite ill, so won't be able to help much with this (I need to get better to actually make it to the conference at all). Cheers, Carl Friedrich From lac at openend.se Sun Jun 21 19:59:46 2009 From: lac at openend.se (Laura Creighton) Date: Sun, 21 Jun 2009 19:59:46 +0200 Subject: [pypy-dev] Europython slides In-Reply-To: Message from Carl Friedrich Bolz of "Sun, 21 Jun 2009 17:07:13 +0200." <4A3E4CA1.7060207@gmx.de> References: <4A3E4CA1.7060207@gmx.de> Message-ID: <200906211759.n5LHxkVv028221@theraft.openend.se> In a message of Sun, 21 Jun 2009 17:07:13 +0200, Carl Friedrich Bolz writes: >Hi all, > >just wanted to remind that we are asked to submit the Europython slides >in advance, I think the date was March 22nd (?). Unfortunately I am >quite ill, so won't be able to help much with this (I need to get better >to actually make it to the conference at all). > >Cheers, > >Carl Friedrich 1. Get Better. 2. You meant July, I think. :) 3. Really Get Better. 4, We'd like the slides as soon as we can get them ... Laura From cfbolz at gmx.de Sun Jun 21 20:21:56 2009 From: cfbolz at gmx.de (Carl Friedrich Bolz) Date: Sun, 21 Jun 2009 20:21:56 +0200 Subject: [pypy-dev] Europython slides In-Reply-To: <200906211759.n5LHxkVv028221@theraft.openend.se> References: <4A3E4CA1.7060207@gmx.de> <200906211759.n5LHxkVv028221@theraft.openend.se> Message-ID: <4A3E7A44.30904@gmx.de> Hi Laura, Laura Creighton wrote: > In a message of Sun, 21 Jun 2009 17:07:13 +0200, Carl Friedrich Bolz writes: >> just wanted to remind that we are asked to submit the Europython slides >> in advance, I think the date was March 22nd (?). Unfortunately I am >> quite ill, so won't be able to help much with this (I need to get better >> to actually make it to the conference at all). > > 1. Get Better. Thanks. I am (not) working on it. > 2. You meant July, I think. :) I think I probably mean June? Otherwise it would be after the conference and not much of a problem :-). Carl Friedrich From lac at openend.se Sun Jun 21 23:22:59 2009 From: lac at openend.se (Laura Creighton) Date: Sun, 21 Jun 2009 23:22:59 +0200 Subject: [pypy-dev] Europython slides In-Reply-To: Message from Carl Friedrich Bolz of "Sun, 21 Jun 2009 20:21:56 +0200." <4A3E7A44.30904@gmx.de> References: <4A3E4CA1.7060207@gmx.de> <200906211759.n5LHxkVv028221@theraft.openend.se> <4A3E7A44.30904@gmx.de> Message-ID: <200906212122.n5LLMxvE006870@theraft.openend.se> In a message of Sun, 21 Jun 2009 20:21:56 +0200, Carl Friedrich Bolz writes: >Hi Laura, > >Laura Creighton wrote: >> In a message of Sun, 21 Jun 2009 17:07:13 +0200, Carl Friedrich Bolz wr >ites: >>> just wanted to remind that we are asked to submit the Europython slide >s >>> in advance, I think the date was March 22nd (?). Unfortunately I am >>> quite ill, so won't be able to help much with this (I need to get bett >er >>> to actually make it to the conference at all). >> >> 1. Get Better. > >Thanks. I am (not) working on it. > >> 2. You meant July, I think. :) > >I think I probably mean June? Otherwise it would be after the conference >and not much of a problem :-). > >Carl Friedrich I think I need a new head. This one is worn out. Get better, Laura From fijall at gmail.com Fri Jun 26 21:41:24 2009 From: fijall at gmail.com (Maciej Fijalkowski) Date: Fri, 26 Jun 2009 13:41:24 -0600 Subject: [pypy-dev] compcache on bigdog-vm2 Message-ID: <693bc9ab0906261241u357839cdv5bad22936ac922d7@mail.gmail.com> Hey. I installed compcache on bigdog-vm2 to increase it's ram/swap size. Details are here: http://code.google.com/p/compcache/wiki/CompilingAndUsing Please report if anything is misbehaving. Cheers, fijal