[matplotlib-devel] segfault on specgram in svn
Hello, I've just switched to Python 2.5 and at the same time upgraded to numpy 1.0.1 with today's svn matplotlib, using the QtAgg backend (PyQt3 3.17). This is on an AMD64 (Opteron) machine. I get a segfault after these operations: In [1]:import numpy as N In [2]:specgram( N.random.randn( 256*500 ) ) Segmentation fault (core dumped) I don't know if this will help, but here is the backtrace: (gdb) bt #0 0x2aaab37e2ac8 in Image::flipud_out (this=0xf9bf10, [EMAIL PROTECTED]) at _image.cpp:110 #1 0x2aaab3829df4 in RendererAgg::draw_image (this=0xf04980, [EMAIL PROTECTED]) at _ns_backend_agg.cpp:1994 #2 0x2aaab3840367 in Py::PythonExtension::method_varargs_call_handler (_self_and_name_tuple=, _args=0x2aaab397bdb8) at Extensions.hxx:683 #3 0x2ac826dc in PyEval_EvalFrameEx (f=0xf76550, throwflag=) at ceval.c:3566 #4 0x2ac82e00 in PyEval_EvalCodeEx (co=0x2aaab34115d0, globals=, locals=, args=0xf764c8, argcount=2, kws=0xf764d8, kwcount=0, defs=0x0, defcount=0, closure=0x0) at ceval.c:2833 #5 0x2ac81cdc in PyEval_EvalFrameEx (f=0xf762c0, throwflag=) at ceval.c:3662 #6 0x2ac82e00 in PyEval_EvalCodeEx (co=0x2aaab2a35648, globals=, locals=, args=0x2, argcount=2, kws=0xf08a88, kwcount=0, defs=0x2aaab3679068, defcount=2, closure=0x0) at ceval.c:2833 #7 0x2ac81cdc in PyEval_EvalFrameEx (f=0xf08880, throwflag=) at ceval.c:3662 #8 0x2ac81d79 in PyEval_EvalFrameEx (f=0xf069f0, throwflag=) at ceval.c:3652 #9 0x2ac82e00 in PyEval_EvalCodeEx (co=0x2aaab37a5a08, globals=, locals=, args=0x2aaab39641e8, argcount=1, kws=0x0, kwcount=0, defs=0x0, defcount=0, closure=0x0) at ceval.c:2833 #10 0x2ac21779 in function_call (func=0x2aaab3708938, arg=0x2aaab39641d0, kw=0x0) at funcobject.c:517 #11 0x2abff8c1 in PyObject_Call (func=, arg=, kw=) at abstract.c:1860 #12 0x2ac08144 in instancemethod_call (func=, arg=0x2aaab39641d0, kw=0x0) at classobject.c:2493 #13 0x2abff8c1 in PyObject_Call (func=, arg=, kw=) at abstract.c:1860 #14 0x2ac80f1e in PyEval_EvalFrameEx (f=0xf06620, throwflag=) at ceval.c:3777 #15 0x2ac82e00 in PyEval_EvalCodeEx (co=0x2d6e3c60, globals=, locals=, args=0x2aaab3715c38, argcount=2, kws=0x0, kwcount=0, defs=0x0, defcount=0, closure=0x0) at ceval.c:2833 #16 0x2ac21779 in function_call (func=0x2aaab3728230, arg=0x2aaab3715c20, kw=0x0) at funcobject.c:517 #17 0x2abff8c1 in PyObject_Call (func=, arg=, kw=) at abstract.c:1860 #18 0x2ac08144 in instancemethod_call (func=, arg=0x2aaab3715c20, kw=0x0) at classobject.c:2493 #19 0x2abff8c1 in PyObject_Call (func=, arg=, kw=) at abstract.c:1860 I don't get the same error on a P4 machine with the same setup except using Python 2.4. For some reason I thought that matplotlib supported 2.5, but then I started looking and couldn't see that documented anywhere. Does it? Thanks, Glen Mabey - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV ___ Matplotlib-devel mailing list Matplotlib-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
Re: [matplotlib-devel] segfault on specgram in svn
> "Glen" == Glen W Mabey <[EMAIL PROTECTED]> writes: Glen> Hello, I've just switched to Python 2.5 and at the same time Glen> upgraded to numpy 1.0.1 with today's svn matplotlib, using Glen> the QtAgg backend (PyQt3 3.17). This is on an AMD64 Glen> (Opteron) machine. Glen> I get a segfault after these operations: Glen> In [1]:import numpy as N In [2]:specgram( N.random.randn( Glen> 256*500 ) ) Segmentation fault (core dumped) 1) Are you sure that matplotlib's numerix setting is numpy? 2) Did you do a *clean* build of mpl: ie > sudo rm -rf build > sudo python setup.py install The latter is very important when upgrading Numeric/numpy/numarray It could be an AMD64 bit problem, though... JDH - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV ___ Matplotlib-devel mailing list Matplotlib-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
Re: [matplotlib-devel] segfault on specgram in svn
On Fri, Dec 08, 2006 at 01:44:03PM -0600, John Hunter wrote: > > "Glen" == Glen W Mabey <[EMAIL PROTECTED]> writes: > > Glen> Hello, I've just switched to Python 2.5 and at the same time > Glen> upgraded to numpy 1.0.1 with today's svn matplotlib, using > Glen> the QtAgg backend (PyQt3 3.17). This is on an AMD64 > Glen> (Opteron) machine. > > Glen> I get a segfault after these operations: > > Glen> In [1]:import numpy as N In [2]:specgram( N.random.randn( > Glen> 256*500 ) ) Segmentation fault (core dumped) > > 1) Are you sure that matplotlib's numerix setting is numpy? Yep. It has been for a long time. > 2) Did you do a *clean* build of mpl: ie >> sudo rm -rf build >> sudo python setup.py install I'm pretty sure it was clean, because I upgraded to the svn version at the same time. I'm rebuilding it now, though, just to make sure, and I'll post if there is any difference in result. Thanks for your suggestions. Glen - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV ___ Matplotlib-devel mailing list Matplotlib-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
Re: [matplotlib-devel] segfault on specgram in svn
On Fri, Dec 08, 2006 at 01:54:56PM -0600, Glen W. Mabey wrote: > On Fri, Dec 08, 2006 at 01:44:03PM -0600, John Hunter wrote: > > > "Glen" == Glen W Mabey <[EMAIL PROTECTED]> writes: > > > > Glen> Hello, I've just switched to Python 2.5 and at the same time > > Glen> upgraded to numpy 1.0.1 with today's svn matplotlib, using > > Glen> the QtAgg backend (PyQt3 3.17). This is on an AMD64 > > Glen> (Opteron) machine. > > > > Glen> I get a segfault after these operations: > > > > Glen> In [1]:import numpy as N In [2]:specgram( N.random.randn( > > Glen> 256*500 ) ) Segmentation fault (core dumped) > > > > 1) Are you sure that matplotlib's numerix setting is numpy? > > Yep. It has been for a long time. > > > 2) Did you do a *clean* build of mpl: ie > >> sudo rm -rf build > >> sudo python setup.py install > > I'm pretty sure it was clean, because I upgraded to the svn version at > the same time. I'm rebuilding it now, though, just to make sure, and > I'll post if there is any difference in result. Okay, it's just my fault. Turns out there is only a segfault when I include a patch I'm working on ... Thanks again, Glen - Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys - and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV ___ Matplotlib-devel mailing list Matplotlib-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/matplotlib-devel
