Menu
▾
▴
pysparse-developers
|
From: Roman G. <rom...@gm...> - 2010-01-26 21:28:55
|
Hello Triggered by a recent email on the pysparse-users list I tried to compile pysparse 1.1 for Windows and Python 2.6. I used Visual C++ (Express Edition) and the latest ACML for BLAS and LAPACK (both are available for free). Except for the superlu package (which I skipped) only a few changes to the source code were necessary (mostly due to the lack of C99 support in the MS C compiler) to build the modules. Unfortunately one of the unit tests in spmatrix_test failed, but basically it seemed to work. (Later I saw that the same unit test also fails on Linux, so it might be a general problem.) I have two questions: Do you know of anybody who has successfully built pysparse using Visual C++ recently? Do you think it would be desirable to be able to build using Visual C++ or is build process using MinGW straight-forward and sufficient? Best regards, Roman |
|
From: Roman G. <rom...@gm...> - 2010-02-03 23:01:14
|
Hi Just an update... I'm still trying to build pysparse in Windows using VC++ and ACML for Python 2.6. Basically it seems to be working now, but only if I link ACML dynamically (as DLL), which not suitable for creating an installer. If I link the static ACML I still get dependencies to Fortran-related DLLs. I hope I find a solution soon. Regards, Roman |
|
From: Daniel W. <dan...@gm...> - 2010-02-04 15:22:51
|
Good luck Roman. I'm trying to build it on Windows 7 running as a virtual machine. It took me a day to figure out that I needed to run the "command line application" as an administrator before it does anything! On Wed, Feb 3, 2010 at 6:01 PM, Roman Geus <rom...@gm...> wrote: > Hi > > Just an update... > > I'm still trying to build pysparse in Windows using VC++ and ACML for > Python 2.6. Basically it seems to be working now, but only if I link > ACML dynamically (as DLL), which not suitable for creating an installer. > > If I link the static ACML I still get dependencies to Fortran-related DLLs. > > I hope I find a solution soon. > > Regards, > Roman > > > > > > ------------------------------------------------------------------------------ > The Planet: dedicated and managed hosting, cloud storage, colocation > Stay online with enterprise data centers and the best network in the business > Choose flexible plans and management services without long-term contracts > Personal 24x7 support from experience hosting pros just a phone call away. > http://p.sf.net/sfu/theplanet-com > _______________________________________________ > Pysparse-developers mailing list > Pys...@li... > https://lists.sourceforge.net/lists/listinfo/pysparse-developers > -- Daniel Wheeler |
|
From: Roman G. <rom...@gm...> - 2010-02-07 17:04:32
|
Hello Daniel and Dominique Finally I managed to create a Windows installer for Python 2.6 and Numpy 1.4 based on the current svn trunk available at http://ovh.dl.sourceforge.net/project/pysparse/devel/pysparse-1.2a1.win32-py2.6.exe (BTW: there must be a better way to share files between developers...) It was built using VC++ 2008 and ACML. I have commit the necessary code changes and some install documentation. Is anybody willing to test it? Best regards, Roman On 04.02.2010 16:22, Daniel Wheeler wrote: > Good luck Roman. I'm trying to build it on Windows 7 running as a > virtual machine. It took me a day to figure out that I needed to run > the "command line application" as an administrator before it does > anything! > > On Wed, Feb 3, 2010 at 6:01 PM, Roman Geus<rom...@gm...> wrote: > >> Hi >> >> Just an update... >> >> I'm still trying to build pysparse in Windows using VC++ and ACML for >> Python 2.6. Basically it seems to be working now, but only if I link >> ACML dynamically (as DLL), which not suitable for creating an installer. >> >> If I link the static ACML I still get dependencies to Fortran-related DLLs. >> >> I hope I find a solution soon. >> >> Regards, >> Roman >> >> >> >> >> >> ------------------------------------------------------------------------------ >> The Planet: dedicated and managed hosting, cloud storage, colocation >> Stay online with enterprise data centers and the best network in the business >> Choose flexible plans and management services without long-term contracts >> Personal 24x7 support from experience hosting pros just a phone call away. >> http://p.sf.net/sfu/theplanet-com >> _______________________________________________ >> Pysparse-developers mailing list >> Pys...@li... >> https://lists.sourceforge.net/lists/listinfo/pysparse-developers >> >> > > > |
|
From: Roman G. <rom...@gm...> - 2010-02-07 17:48:04
|
Hi again If you are interested in downloading the new Windows installer please use the following URL: http://hamsel.users.sourceforge.net/pysparse-1.2a1.win32-py2.6.exe I will delete the installer from the list of public files, so the old URL becomes invalid. Cheers, Roman On 07.02.2010 18:04, Roman Geus wrote: > Hello Daniel and Dominique > > Finally I managed to create a Windows installer for Python 2.6 and > Numpy 1.4 based on the current svn trunk available at > http://ovh.dl.sourceforge.net/project/pysparse/devel/pysparse-1.2a1.win32-py2.6.exe > > > (BTW: there must be a better way to share files between developers...) > > It was built using VC++ 2008 and ACML. > > I have commit the necessary code changes and some install documentation. > > Is anybody willing to test it? > > Best regards, > Roman > > On 04.02.2010 16:22, Daniel Wheeler wrote: >> Good luck Roman. I'm trying to build it on Windows 7 running as a >> virtual machine. It took me a day to figure out that I needed to run >> the "command line application" as an administrator before it does >> anything! >> >> On Wed, Feb 3, 2010 at 6:01 PM, Roman Geus<rom...@gm...> wrote: >>> Hi >>> >>> Just an update... >>> >>> I'm still trying to build pysparse in Windows using VC++ and ACML for >>> Python 2.6. Basically it seems to be working now, but only if I link >>> ACML dynamically (as DLL), which not suitable for creating an >>> installer. >>> >>> If I link the static ACML I still get dependencies to >>> Fortran-related DLLs. >>> >>> I hope I find a solution soon. >>> >>> Regards, >>> Roman >>> >>> >>> >>> >>> >>> ------------------------------------------------------------------------------ >>> >>> The Planet: dedicated and managed hosting, cloud storage, colocation >>> Stay online with enterprise data centers and the best network in the >>> business >>> Choose flexible plans and management services without long-term >>> contracts >>> Personal 24x7 support from experience hosting pros just a phone call >>> away. >>> http://p.sf.net/sfu/theplanet-com >>> _______________________________________________ >>> Pysparse-developers mailing list >>> Pys...@li... >>> https://lists.sourceforge.net/lists/listinfo/pysparse-developers >>> >> >> > |
|
From: Daniel W. <dan...@gm...> - 2010-02-11 16:25:42
|
I installed this on Windows 7 virtual machine. When I run "import pysparse" in the python command prompt, I get the following error message in a dialogue box "python.exe has stopped working". No idea what is going on here or why a dialogue box pops up when it's importing a python script. Any ideas? On Sun, Feb 7, 2010 at 12:47 PM, Roman Geus <rom...@gm...> wrote: > Hi again > > If you are interested in downloading the new Windows installer please > use the following URL: > http://hamsel.users.sourceforge.net/pysparse-1.2a1.win32-py2.6.exe > > I will delete the installer from the list of public files, so the old > URL becomes invalid. > > Cheers, > Roman > > On 07.02.2010 18:04, Roman Geus wrote: >> Hello Daniel and Dominique >> >> Finally I managed to create a Windows installer for Python 2.6 and >> Numpy 1.4 based on the current svn trunk available at >> http://ovh.dl.sourceforge.net/project/pysparse/devel/pysparse-1.2a1.win32-py2.6.exe >> >> >> (BTW: there must be a better way to share files between developers...) >> >> It was built using VC++ 2008 and ACML. >> >> I have commit the necessary code changes and some install documentation. >> >> Is anybody willing to test it? >> >> Best regards, >> Roman >> >> On 04.02.2010 16:22, Daniel Wheeler wrote: >>> Good luck Roman. I'm trying to build it on Windows 7 running as a >>> virtual machine. It took me a day to figure out that I needed to run >>> the "command line application" as an administrator before it does >>> anything! >>> >>> On Wed, Feb 3, 2010 at 6:01 PM, Roman Geus<rom...@gm...> wrote: >>>> Hi >>>> >>>> Just an update... >>>> >>>> I'm still trying to build pysparse in Windows using VC++ and ACML for >>>> Python 2.6. Basically it seems to be working now, but only if I link >>>> ACML dynamically (as DLL), which not suitable for creating an >>>> installer. >>>> >>>> If I link the static ACML I still get dependencies to >>>> Fortran-related DLLs. >>>> >>>> I hope I find a solution soon. >>>> >>>> Regards, >>>> Roman >>>> >>>> >>>> >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> >>>> The Planet: dedicated and managed hosting, cloud storage, colocation >>>> Stay online with enterprise data centers and the best network in the >>>> business >>>> Choose flexible plans and management services without long-term >>>> contracts >>>> Personal 24x7 support from experience hosting pros just a phone call >>>> away. >>>> http://p.sf.net/sfu/theplanet-com >>>> _______________________________________________ >>>> Pysparse-developers mailing list >>>> Pys...@li... >>>> https://lists.sourceforge.net/lists/listinfo/pysparse-developers >>>> >>> >>> >> > > > ------------------------------------------------------------------------------ > The Planet: dedicated and managed hosting, cloud storage, colocation > Stay online with enterprise data centers and the best network in the business > Choose flexible plans and management services without long-term contracts > Personal 24x7 support from experience hosting pros just a phone call away. > http://p.sf.net/sfu/theplanet-com > _______________________________________________ > Pysparse-developers mailing list > Pys...@li... > https://lists.sourceforge.net/lists/listinfo/pysparse-developers > -- Daniel Wheeler |
|
From: Roman G. <rom...@gm...> - 2010-02-11 20:39:30
|
Hi Daniel Thanks for trying. I have only tested it on two Windows XP machines. It could be missing DLLs: The following need to be available: python26.dll (from the Python installation) msvcr90.dll (from the OS) kernel32.dll (from the OS) imagehlp.dll (from the OS) libifcoremd.dll (from pysparse) libmmd.dll (from pysparse) Unfortunately I'm not a Windows expert and I have no Windows7 available for testing. So it could something entirely different. Hopefully the Python modules compiled with mingw will work better. Regards, Roman On 11.02.2010 17:25, Daniel Wheeler wrote: > I installed this on Windows 7 virtual machine. When I run "import > pysparse" in the python command prompt, I get the following error > message in a dialogue box "python.exe has stopped working". No idea > what is going on here or why a dialogue box pops up when it's > importing a python script. Any ideas? > > On Sun, Feb 7, 2010 at 12:47 PM, Roman Geus<rom...@gm...> wrote: > >> Hi again >> >> If you are interested in downloading the new Windows installer please >> use the following URL: >> http://hamsel.users.sourceforge.net/pysparse-1.2a1.win32-py2.6.exe >> >> I will delete the installer from the list of public files, so the old >> URL becomes invalid. >> >> Cheers, >> Roman >> >> On 07.02.2010 18:04, Roman Geus wrote: >> >>> Hello Daniel and Dominique >>> >>> Finally I managed to create a Windows installer for Python 2.6 and >>> Numpy 1.4 based on the current svn trunk available at >>> http://ovh.dl.sourceforge.net/project/pysparse/devel/pysparse-1.2a1.win32-py2.6.exe >>> >>> >>> (BTW: there must be a better way to share files between developers...) >>> >>> It was built using VC++ 2008 and ACML. >>> >>> I have commit the necessary code changes and some install documentation. >>> >>> Is anybody willing to test it? >>> >>> Best regards, >>> Roman >>> >>> On 04.02.2010 16:22, Daniel Wheeler wrote: >>> >>>> Good luck Roman. I'm trying to build it on Windows 7 running as a >>>> virtual machine. It took me a day to figure out that I needed to run >>>> the "command line application" as an administrator before it does >>>> anything! >>>> >>>> On Wed, Feb 3, 2010 at 6:01 PM, Roman Geus<rom...@gm...> wrote: >>>> >>>>> Hi >>>>> >>>>> Just an update... >>>>> >>>>> I'm still trying to build pysparse in Windows using VC++ and ACML for >>>>> Python 2.6. Basically it seems to be working now, but only if I link >>>>> ACML dynamically (as DLL), which not suitable for creating an >>>>> installer. >>>>> >>>>> If I link the static ACML I still get dependencies to >>>>> Fortran-related DLLs. >>>>> >>>>> I hope I find a solution soon. >>>>> >>>>> Regards, >>>>> Roman >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> ------------------------------------------------------------------------------ >>>>> >>>>> The Planet: dedicated and managed hosting, cloud storage, colocation >>>>> Stay online with enterprise data centers and the best network in the >>>>> business >>>>> Choose flexible plans and management services without long-term >>>>> contracts >>>>> Personal 24x7 support from experience hosting pros just a phone call >>>>> away. >>>>> http://p.sf.net/sfu/theplanet-com >>>>> _______________________________________________ >>>>> Pysparse-developers mailing list >>>>> Pys...@li... >>>>> https://lists.sourceforge.net/lists/listinfo/pysparse-developers >>>>> >>>>> >>>> >>>> >>> >> >> ------------------------------------------------------------------------------ >> The Planet: dedicated and managed hosting, cloud storage, colocation >> Stay online with enterprise data centers and the best network in the business >> Choose flexible plans and management services without long-term contracts >> Personal 24x7 support from experience hosting pros just a phone call away. >> http://p.sf.net/sfu/theplanet-com >> _______________________________________________ >> Pysparse-developers mailing list >> Pys...@li... >> https://lists.sourceforge.net/lists/listinfo/pysparse-developers >> >> > > > |
|
From: Roman G. <rom...@gm...> - 2010-02-28 21:30:22
|
Hello Daniel I have tested the new Windows installer on a Windows 7 Enterprise 32-bit system (fresh install) and had no particular problems with it. Did you test using a 64-bit edition? However there seems to be an issue with the superlu module: The tests fail on Windows 7 and also on Windows XP. I'll look into that. Regards, Roman On 11.02.2010 17:25, Daniel Wheeler wrote: > I installed this on Windows 7 virtual machine. When I run "import > pysparse" in the python command prompt, I get the following error > message in a dialogue box "python.exe has stopped working". No idea > what is going on here or why a dialogue box pops up when it's > importing a python script. Any ideas? > > On Sun, Feb 7, 2010 at 12:47 PM, Roman Geus<rom...@gm...> wrote: > >> Hi again >> >> If you are interested in downloading the new Windows installer please >> use the following URL: >> http://hamsel.users.sourceforge.net/pysparse-1.2a1.win32-py2.6.exe >> >> I will delete the installer from the list of public files, so the old >> URL becomes invalid. >> >> Cheers, >> Roman >> >> On 07.02.2010 18:04, Roman Geus wrote: >> >>> Hello Daniel and Dominique >>> >>> Finally I managed to create a Windows installer for Python 2.6 and >>> Numpy 1.4 based on the current svn trunk available at >>> http://ovh.dl.sourceforge.net/project/pysparse/devel/pysparse-1.2a1.win32-py2.6.exe >>> >>> >>> (BTW: there must be a better way to share files between developers...) >>> >>> It was built using VC++ 2008 and ACML. >>> >>> I have commit the necessary code changes and some install documentation. >>> >>> Is anybody willing to test it? >>> >>> Best regards, >>> Roman >>> >>> On 04.02.2010 16:22, Daniel Wheeler wrote: >>> >>>> Good luck Roman. I'm trying to build it on Windows 7 running as a >>>> virtual machine. It took me a day to figure out that I needed to run >>>> the "command line application" as an administrator before it does >>>> anything! >>>> >>>> On Wed, Feb 3, 2010 at 6:01 PM, Roman Geus<rom...@gm...> wrote: >>>> >>>>> Hi >>>>> >>>>> Just an update... >>>>> >>>>> I'm still trying to build pysparse in Windows using VC++ and ACML for >>>>> Python 2.6. Basically it seems to be working now, but only if I link >>>>> ACML dynamically (as DLL), which not suitable for creating an >>>>> installer. >>>>> >>>>> If I link the static ACML I still get dependencies to >>>>> Fortran-related DLLs. >>>>> >>>>> I hope I find a solution soon. >>>>> >>>>> Regards, >>>>> Roman >>>>> >>>>> >>>>> >>>>> >>>>> >>>>> ------------------------------------------------------------------------------ >>>>> >>>>> The Planet: dedicated and managed hosting, cloud storage, colocation >>>>> Stay online with enterprise data centers and the best network in the >>>>> business >>>>> Choose flexible plans and management services without long-term >>>>> contracts >>>>> Personal 24x7 support from experience hosting pros just a phone call >>>>> away. >>>>> http://p.sf.net/sfu/theplanet-com >>>>> _______________________________________________ >>>>> Pysparse-developers mailing list >>>>> Pys...@li... >>>>> https://lists.sourceforge.net/lists/listinfo/pysparse-developers >>>>> >>>>> >>>> >>>> >>> >> >> ------------------------------------------------------------------------------ >> The Planet: dedicated and managed hosting, cloud storage, colocation >> Stay online with enterprise data centers and the best network in the business >> Choose flexible plans and management services without long-term contracts >> Personal 24x7 support from experience hosting pros just a phone call away. >> http://p.sf.net/sfu/theplanet-com >> _______________________________________________ >> Pysparse-developers mailing list >> Pys...@li... >> https://lists.sourceforge.net/lists/listinfo/pysparse-developers >> >> > > > |
|
From: Daniel W. <dan...@gm...> - 2010-03-01 14:46:15
|
On Sun, Feb 28, 2010 at 4:30 PM, Roman Geus <rom...@gm...> wrote: > Hello Daniel > > I have tested the new Windows installer on a Windows 7 Enterprise 32-bit > system (fresh install) and had no particular problems with it. Did you test > using a 64-bit edition? No, it was 32. I haven't tried since you gave me the advice about the dlls. Maybe put it up on the sourceforge site and see if anyone else complains. I seem to recall you built the installer from the latest version of trunk. We really need a python 2.6 installer for version 1.1 rather than trunk. Anyway, as you see fit. > However there seems to be an issue with the superlu module: The tests fail > on Windows 7 and also on Windows XP. I'll look into that. If I get another chance, I'll give it a shot, but I never seem to make time to do these sorts of things these days. Cheers -- Daniel Wheeler |
|
From: Roman G. <rom...@gm...> - 2010-03-01 17:21:31
|
Hi Daniel On 01.03.2010, at 15:46, Daniel Wheeler <dan...@gm...> wrote: > On Sun, Feb 28, 2010 at 4:30 PM, Roman Geus <rom...@gm...> > wrote: >> Hello Daniel >> >> I have tested the new Windows installer on a Windows 7 Enterprise >> 32-bit >> system (fresh install) and had no particular problems with it. Did >> you test >> using a 64-bit edition? > > No, it was 32. I haven't tried since you gave me the advice about the > dlls. Maybe put it up on the sourceforge site and see if anyone else > complains. There are still problems in the superlu module in my installer. Those have to probably be sorted out before it can be released. > I seem to recall you built the installer from the latest > version of trunk. We really need a python 2.6 installer for version > 1.1 rather than trunk. Anyway, as you see fit. There were some modifications necessary to make the source compatible with MSVC compilers - since there is no 1.1 branch, I had to work in the trunk. We can look into that when the rest works. > >> However there seems to be an issue with the superlu module: The >> tests fail >> on Windows 7 and also on Windows XP. I'll look into that. > > If I get another chance, I'll give it a shot, but I never seem to make > time to do these sorts of things these days. I know... Regards Roman > > Cheers > > -- > Daniel Wheeler > > --- > --- > --- > --------------------------------------------------------------------- > Download Intel® Parallel Studio Eval > Try the new software tools for yourself. Speed compiling, find bugs > proactively, and fine-tune applications for parallel performance. > See why Intel Parallel Studio got high marks during beta. > http://p.sf.net/sfu/intel-sw-dev > _______________________________________________ > Pysparse-developers mailing list > Pys...@li... > https://lists.sourceforge.net/lists/listinfo/pysparse-developers |
|
From: Daniel W. <dan...@gm...> - 2010-04-06 14:32:06
|
Hi, I've been struggling with trying to build pysparse version 1.1 for python 2.6 on windows, both XP and 7. I'm using the mingw that comes with pythonxy. I compiled blas and lapack from the source. I had to go back a few versions of lapack in order for it to work with the gcc (gfortran) that comes with pythonxy at least that seemed to work. However, a whole bunch of compile flags get automatically added to the pysparse compile step by mingw and I have no idea what they do. Anyway, the long and the short is that I can build pysparse, but I can't import any of the compiled modules without either getting a DLL error (if lapack.dll and blas.dll are not in the DLL directory) or a memory error (if lapack.dll and blas.dll are in the DLL directory). The weird thing is that I need both the .lib (or .a) versions as well as the .dll versions of lapack and blas. I realized that I'd stashed liblapack.a and libblas.a in the tools/win32 of the pysparse dist for safe keeping. Anyway, I get the DLL error when I use those because I don't have the dll versions of those libraries. I need to figure out how to just include the blas and lapack statically using pythonxy's mingw. This wasn't a problem in previous releases of pythonxy. Anyway, I need to start asking questions on the pythonxy list. In the mean time, is it possible that someone could get a python 2.6, version 1.1 installer up (using a clean checkout of tags/version-1_1). Thanks. On Mon, Mar 1, 2010 at 1:21 PM, Roman Geus <rom...@gm...> wrote: > There were some modifications necessary to make the source compatible with > MSVC compilers - since there is no 1.1 branch, I had to work in the trunk. > We can look into that when the rest works. Just use tags. If things need to be fixed you can create a branch from tags/version-1_1 (to branches/version-1_1). Tags should just be place holders. -- Daniel Wheeler |
|
From: Roman G. <rom...@gm...> - 2010-04-06 20:45:57
|
Hi Daniel On 06.04.2010 16:31, Daniel Wheeler wrote: > Hi, I've been struggling with trying to build pysparse version 1.1 for > python 2.6 on windows, both XP and 7. I'm using the mingw that comes > with pythonxy. I compiled blas and lapack from the source. I had to go > back a few versions of lapack in order for it to work with the gcc > (gfortran) that comes with pythonxy at least that seemed to work. > However, a whole bunch of compile flags get automatically added to the > pysparse compile step by mingw and I have no idea what they do. > Anyway, the long and the short is that I can build pysparse, but I > can't import any of the compiled modules without either getting a DLL > error (if lapack.dll and blas.dll are not in the DLL directory) or a > memory error (if lapack.dll and blas.dll are in the DLL directory). > > The weird thing is that I need both the .lib (or .a) versions as well > as the .dll versions of lapack and blas. > To my knowledge that's normal under Windows: if you link dynamically you need a .lib file (at compile time) and a .dll file. If you link statically you need only .lib files. > I realized that I'd stashed liblapack.a and libblas.a in the > tools/win32 of the pysparse dist for safe keeping. Anyway, I get the > DLL error when I use those because I don't have the dll versions of > those libraries. I need to figure out how to just include the blas and > lapack statically using pythonxy's mingw. This wasn't a problem in > previous releases of pythonxy. Anyway, I need to start asking > questions on the pythonxy list. > > In the mean time, is it possible that someone could get a python 2.6, > version 1.1 installer up (using a clean checkout of tags/version-1_1). > Thanks. > > On Mon, Mar 1, 2010 at 1:21 PM, Roman Geus<rom...@gm...> wrote: > > > >> There were some modifications necessary to make the source compatible with >> MSVC compilers - since there is no 1.1 branch, I had to work in the trunk. >> We can look into that when the rest works. >> > Just use tags. If things need to be fixed you can create a branch from > tags/version-1_1 (to branches/version-1_1). Tags should just be place > holders. > Yes, right. But my windows build still produces wrong results in the superlu module. -- Roman |
|
From: Daniel W. <dan...@gm...> - 2010-04-06 21:00:51
|
On Tue, Apr 6, 2010 at 4:45 PM, Roman Geus <rom...@gm...> wrote: >> Just use tags. If things need to be fixed you can create a branch from >> tags/version-1_1 (to branches/version-1_1). Tags should just be place >> holders. >> > > Yes, right. But my windows build still produces wrong results in the superlu > module. Ho-hum, then we are hamstrung at the moment. I'll try and figure out how to make pythonxy build statically, probably the most useful thing I can do for now. We just released a new version of fipy (version 2.1) so I need to deal with this issue. -- Daniel Wheeler |
|
From: Roman G. <rom...@gm...> - 2010-04-06 22:00:21
|
Hi Daniel I just gave it another try. All I can say is that this Windows/superlu bug is quite serious: even 4-by-4 matrices produce completely wrong results. (1-by-1 matrices work though ;-) I this requires more time to investigate (and the time I can spare is unfortunately limited). Next I will try to call superlu from C just to check whether the problem is related in any way to Python. Regards, Roman On 06.04.2010 23:00, Daniel Wheeler wrote: > On Tue, Apr 6, 2010 at 4:45 PM, Roman Geus<rom...@gm...> wrote: > >>> Just use tags. If things need to be fixed you can create a branch from >>> tags/version-1_1 (to branches/version-1_1). Tags should just be place >>> holders. >>> >>> >> Yes, right. But my windows build still produces wrong results in the superlu >> module. >> > Ho-hum, then we are hamstrung at the moment. I'll try and figure out > how to make pythonxy build statically, probably the most useful thing > I can do for now. We just released a new version of fipy (version 2.1) > so I need to deal with this issue. > > |
|
From: Roman G. <rom...@gm...> - 2010-04-12 21:47:44
|
Hi all FYI: I found a problem in the Windows build of the SuperLU module that caused the wrong results. The uppercase symbols of the double-precision BLAS routines (which are used in Windows) were mapped to single-precision BLAS routines (in Cnames.h). Next, I work on the installer for Win32 and pysparse 1.1. Let's hope that was the last serious Windows-related issue... Cheers, Roman On 06.04.2010 23:00, Daniel Wheeler wrote: > On Tue, Apr 6, 2010 at 4:45 PM, Roman Geus<rom...@gm...> wrote: > >>> Just use tags. If things need to be fixed you can create a branch from >>> tags/version-1_1 (to branches/version-1_1). Tags should just be place >>> holders. >>> >>> >> Yes, right. But my windows build still produces wrong results in the superlu >> module. >> > Ho-hum, then we are hamstrung at the moment. I'll try and figure out > how to make pythonxy build statically, probably the most useful thing > I can do for now. We just released a new version of fipy (version 2.1) > so I need to deal with this issue. > > |
|
From: Roman G. <rom...@gm...> - 2010-04-13 21:45:47
|
Hi all A releases candidate of the pysparse-1.1 Windows installer for Python 2.6 is now available in the SF file download area. Feedback is welcome. -- Roman PS: There is also a new branch "1_1_x" for maintaining this release. On 12.04.2010 23:47, Roman Geus wrote: > Hi all > > FYI: I found a problem in the Windows build of the SuperLU module that > caused the wrong results. The uppercase symbols of the > double-precision BLAS routines (which are used in Windows) were mapped > to single-precision BLAS routines (in Cnames.h). > > Next, I work on the installer for Win32 and pysparse 1.1. Let's hope > that was the last serious Windows-related issue... > > Cheers, > Roman > > On 06.04.2010 23:00, Daniel Wheeler wrote: >> On Tue, Apr 6, 2010 at 4:45 PM, Roman Geus<rom...@gm...> wrote: >>>> Just use tags. If things need to be fixed you can create a branch from >>>> tags/version-1_1 (to branches/version-1_1). Tags should just be place >>>> holders. >>>> >>> Yes, right. But my windows build still produces wrong results in the >>> superlu >>> module. >> Ho-hum, then we are hamstrung at the moment. I'll try and figure out >> how to make pythonxy build statically, probably the most useful thing >> I can do for now. We just released a new version of fipy (version 2.1) >> so I need to deal with this issue. >> > |
|
From: Dominique O. <dom...@gm...> - 2010-01-26 23:42:32
|
On Tue, Jan 26, 2010 at 9:28 PM, Roman Geus <rom...@gm...> wrote: > Hello > > Triggered by a recent email on the pysparse-users list I tried to > compile pysparse 1.1 for Windows and Python 2.6. > > I used Visual C++ (Express Edition) and the latest ACML for BLAS and > LAPACK (both are available for free). > > Except for the superlu package (which I skipped) only a few changes to > the source code were necessary (mostly due to the lack of C99 support in > the MS C compiler) to build the modules. > > Unfortunately one of the unit tests in spmatrix_test failed, but > basically it seemed to work. (Later I saw that the same unit test also > fails on Linux, so it might be a general problem.) > > I have two questions: > > Do you know of anybody who has successfully built pysparse using Visual > C++ recently? > > Do you think it would be desirable to be able to build using Visual C++ > or is build process using MinGW straight-forward and sufficient? > > Best regards, > Roman I will let Windows users comment on what would best suit their needs. In the meantime I commented out the test that fails. The reason is that when I implemented fancy indexing (which includes slicing), I decided that taking a slice on a matrix would result by default in a non-symmetric matrix, even if the original matrix was symmetric. Clearly, only symmetric slices would yield a symmetric submatrix (although special cases may occur but would be costly to check for). Currently, Pysparse doesn't check for symmetric slices. -- Dominique |
|
From: Roman G. <rom...@gm...> - 2010-01-27 22:09:33
|
Hi Dominique Thanks for the explanation. Maybe it would be feasible to store off-diagonal elements twice (above and below the diagonal) when extracting a submatrix from a symmetric matrix. Regards, Roman On 27.01.2010 00:42, Dominique Orban wrote: > On Tue, Jan 26, 2010 at 9:28 PM, Roman Geus<rom...@gm...> wrote: > >> Hello >> >> Triggered by a recent email on the pysparse-users list I tried to >> compile pysparse 1.1 for Windows and Python 2.6. >> >> I used Visual C++ (Express Edition) and the latest ACML for BLAS and >> LAPACK (both are available for free). >> >> Except for the superlu package (which I skipped) only a few changes to >> the source code were necessary (mostly due to the lack of C99 support in >> the MS C compiler) to build the modules. >> >> Unfortunately one of the unit tests in spmatrix_test failed, but >> basically it seemed to work. (Later I saw that the same unit test also >> fails on Linux, so it might be a general problem.) >> >> I have two questions: >> >> Do you know of anybody who has successfully built pysparse using Visual >> C++ recently? >> >> Do you think it would be desirable to be able to build using Visual C++ >> or is build process using MinGW straight-forward and sufficient? >> >> Best regards, >> Roman >> > I will let Windows users comment on what would best suit their needs. > In the meantime I commented out the test that fails. The reason is > that when I implemented fancy indexing (which includes slicing), I > decided that taking a slice on a matrix would result by default in a > non-symmetric matrix, even if the original matrix was symmetric. > Clearly, only symmetric slices would yield a symmetric submatrix > (although special cases may occur but would be costly to check for). > Currently, Pysparse doesn't check for symmetric slices. > > |
|
From: Roman G. <rom...@gm...> - 2010-01-27 22:15:01
|
Hi Daniel Yes, by all means go ahead with pythonxy. I'll try to get a robust Win32 build using Visual C++. I'll let you know if I succeed. Regards, Roman On 27.01.2010 15:59, Daniel Wheeler wrote: > Hi Roman, I have been using pythonxy as the basis for building > pysparse on windows (switched from enthought, which now has a weird > license). I have also been using some very old compiled versions of > blas and artlas that I really should update. I ran into some problems > upgrading pythonxy to the python 2.6 version so haven't been able to > make a release as yet. I recently installed windows 7 in virtualbox > and I hope to do a clean install of the lastest pythonxy ASAP. As far > as visual c++ is concerned, I have never tried it. I suppose it > doesn't matter what the build process is as long as it works, but I > would rather use the open source tools were available. If you would > like to do it, by all means, but I'll try with pythonxy in the mean > time. Cheers. > > On Tue, Jan 26, 2010 at 4:28 PM, Roman Geus<rom...@gm...> wrote: > >> Hello >> >> Triggered by a recent email on the pysparse-users list I tried to >> compile pysparse 1.1 for Windows and Python 2.6. >> >> I used Visual C++ (Express Edition) and the latest ACML for BLAS and >> LAPACK (both are available for free). >> >> Except for the superlu package (which I skipped) only a few changes to >> the source code were necessary (mostly due to the lack of C99 support in >> the MS C compiler) to build the modules. >> >> Unfortunately one of the unit tests in spmatrix_test failed, but >> basically it seemed to work. (Later I saw that the same unit test also >> fails on Linux, so it might be a general problem.) >> >> I have two questions: >> >> Do you know of anybody who has successfully built pysparse using Visual >> C++ recently? >> >> Do you think it would be desirable to be able to build using Visual C++ >> or is build process using MinGW straight-forward and sufficient? >> >> Best regards, >> Roman >> >> ------------------------------------------------------------------------------ >> The Planet: dedicated and managed hosting, cloud storage, colocation >> Stay online with enterprise data centers and the best network in the business >> Choose flexible plans and management services without long-term contracts >> Personal 24x7 support from experience hosting pros just a phone call away. >> http://p.sf.net/sfu/theplanet-com >> _______________________________________________ >> Pysparse-developers mailing list >> Pys...@li... >> https://lists.sourceforge.net/lists/listinfo/pysparse-developers >> >> > > > |
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X