sage 4.2.1 on 64bit mac 10.6.2 and the plot function crashes

[Aivo Jürgenson]

From: Aivo Jürgenson <aivo.ju...@gmail.com>
Date: November 18, 2009 10:20:05 PM GMT+02:00
To: sage-...@googlegroups.com
Subject: sage 4.2.1 on 64bit mac 10.6.2 and the plot function crashes

Hello.

I'm running the sage 4.2.1 on a 64 bit macosx 10.6.2 and I'm not able to use plot(sin(x),0,4) style functions. It seems that this is related to the http://trac.sagemath.org/sage_trac/ticket/7022 issue, which should be fixed, but unfortunately doesn't work in my machine.

The sage console shows:

2009-11-18 21:53:43+0200 [HTTPChannel,1,127.0.0.1] got EOF subprocess must have crashed...

and the ~/Library/Logs/DiagnosticReports/python_2009-11-18-214233_mind.crash logs are created with the following backtrace:

Thread 0 Crashed: Dispatch queue: com.apple.main-thread
0 libSystem.B.dylib 0x00007fff867f4fe6 __kill + 10
1 libSystem.B.dylib 0x00007fff86895dd3 __abort + 103
2 libSystem.B.dylib 0x00007fff86895e41 abort_report_np + 0
3 libgcc_s.1.dylib 0x0000000103e89d51 uw_init_context_1 + 385
4 libgcc_s.1.dylib 0x0000000103e8a4cc _Unwind_Resume + 76
5 ft2font.so 0x000000010b39c49d FT2Font::FT2Font(std::string) + 4269
6 ft2font.so 0x000000010b39cb28 ft2font_module::new_ft2font(Py::Tuple const&) + 408
7 ft2font.so 0x000000010b39fd82 Py::ExtensionModule<ft2font_module>::invoke_method_varargs(std::string const&, Py::Tuple const&)
+ 162
8 ft2font.so 0x000000010b3ae5d0 method_varargs_call_handler + 512
9 python 0x00000001000b3b25 PyEval_EvalFrameEx + 23061
10 python 0x00000001000b4ce0 PyEval_EvalCodeEx + 2112
11 python 0x00000001000b2afd PyEval_EvalFrameEx + 18925
12 python 0x00000001000b4ce0 PyEval_EvalCodeEx + 2112
13 python 0x0000000100034fb5 function_call + 165
...

Could it be my fault or is it a bug in sage components?

Aivo Jürgenson, aivo.ju...@gmail.com

[William Stein]

2009/11/18 Aivo Jürgenson <aivo.ju...@gmail.com>:

>
> From: Aivo Jürgenson <aivo.ju...@gmail.com>
> Date: November 18, 2009 10:20:05 PM GMT+02:00
> To: sage-...@googlegroups.com
> Subject: sage 4.2.1 on 64bit mac 10.6.2 and the plot function crashes
>
> Hello.
>
> I'm running the sage 4.2.1 on a 64 bit macosx 10.6.2 and I'm not able to use plot(sin(x),0,4) style functions. It seems that this is related to the http://trac.sagemath.org/sage_trac/ticket/7022 issue, which should be fixed, but unfortunately doesn't work in my machine.
>
> The sage console shows:

What happens when you type

sage: plot(sin(x),0,4)

on the command line?

This command definitely works fine for me with 64 bit macosx 10.6.2.

William

>
> 2009-11-18 21:53:43+0200 [HTTPChannel,1,127.0.0.1] got EOF subprocess must have crashed...
>
> and the ~/Library/Logs/DiagnosticReports/python_2009-11-18-214233_mind.crash logs are created with the following backtrace:
>
> Thread 0 Crashed:  Dispatch queue: com.apple.main-thread
> 0   libSystem.B.dylib                   0x00007fff867f4fe6 __kill + 10
> 1   libSystem.B.dylib                   0x00007fff86895dd3 __abort + 103
> 2   libSystem.B.dylib                   0x00007fff86895e41 abort_report_np + 0
> 3   libgcc_s.1.dylib                    0x0000000103e89d51 uw_init_context_1 + 385
> 4   libgcc_s.1.dylib                    0x0000000103e8a4cc _Unwind_Resume + 76
> 5   ft2font.so                          0x000000010b39c49d FT2Font::FT2Font(std::string) + 4269
> 6   ft2font.so                          0x000000010b39cb28 ft2font_module::new_ft2font(Py::Tuple const&) + 408
> 7   ft2font.so                          0x000000010b39fd82 Py::ExtensionModule<ft2font_module>::invoke_method_varargs(std::string const&, Py::Tuple const&)
> + 162
> 8   ft2font.so                          0x000000010b3ae5d0 method_varargs_call_handler + 512
> 9   python                              0x00000001000b3b25 PyEval_EvalFrameEx + 23061
> 10  python                              0x00000001000b4ce0 PyEval_EvalCodeEx + 2112
> 11  python                              0x00000001000b2afd PyEval_EvalFrameEx + 18925
> 12  python                              0x00000001000b4ce0 PyEval_EvalCodeEx + 2112
> 13  python                              0x0000000100034fb5 function_call + 165
> ...
>
> Could it be my fault or is it a bug in sage components?
>
> Aivo Jürgenson, aivo.ju...@gmail.com
>
>
>
>

> --
> To post to this group, send an email to sage-...@googlegroups.com
> To unsubscribe from this group, send an email to sage-devel-...@googlegroups.com
> For more options, visit this group at http://groups.google.com/group/sage-devel
> URL: http://www.sagemath.org



--
William Stein
Associate Professor of Mathematics
University of Washington
http://wstein.org

[Aivo Jürgenson]

> What happens when you type
>
> sage: plot(sin(x),0,4)
>
> on the command line?

>/Applications/sage/sage
----------------------------------------------------------------------
| Sage Version 4.2.1, Release Date: 2009-11-14 |
| Type notebook() for the GUI, and license() for information. |
----------------------------------------------------------------------
WARNING: There is one major unsolved bug in some versions of
Sage on OS X 10.6 that causes an 'Abort trap' crash when
doing certain symbolic computations.
See http://trac.sagemath.org/sage_trac/ticket/7095/.
sage: plot(sin(x),0,4)
/Applications/sage/local/bin/sage-sage: line 203: 20279 Abort
trap sage-ipython "$@" -i

[Ivan Andrus]

FWIW this happens to me too. Running it in sage -gdb I get (the last few lines). Full backtrace at http://math.byu.edu/~gvol/files/sage.log
I'm not sure, but it doesn't seem to be related to 7095. I did a sage -upgrade from a clean build of 4.2.1.alpha0.

/Users/gvol/SageStuff/sage-4.2.1.alpha0/local/lib/python2.6/site-packages/matplotlib/font_manager.pyc in _rebuild()
1290 def _rebuild():
1291 global fontManager
-> 1292 fontManager = FontManager()
1293 pickle_dump(fontManager, _fmcache)
1294 verbose.report("generated new fontManager")

/Users/gvol/SageStuff/sage-4.2.1.alpha0/local/lib/python2.6/site-packages/matplotlib/font_manager.pyc in __init__(self, size, weight)
982 # Load TrueType fonts and create font dictionary.
983
--> 984 self.ttffiles = findSystemFonts(paths) + findSystemFonts()
985
986 for fname in self.ttffiles:

/Users/gvol/SageStuff/sage-4.2.1.alpha0/local/lib/python2.6/site-packages/matplotlib/font_manager.pyc in findSystemFonts(fontpaths, fontext)
339 fontfiles[f] = 1
340
--> 341 for f in get_fontconfig_fonts(fontext):
342 fontfiles[f] = 1
343

/Users/gvol/SageStuff/sage-4.2.1.alpha0/local/lib/python2.6/site-packages/matplotlib/font_manager.pyc in get_fontconfig_fonts(fontext)
301
302 fontfiles = {}
--> 303 status, output = commands.getstatusoutput("fc-list file")
304 if status == 0:
305 for line in output.split('\n'):

/Users/gvol/SageStuff/sage-4.2.1.alpha0/local/lib/python/commands.pyc in getstatusoutput(cmd)
54 import os
55 pipe = os.popen('{ ' + cmd + '; } 2>&1', 'r')
---> 56 text = pipe.read()
57 sts = pipe.close()
58 if sts is None: sts = 0

IOError: [Errno 4] Interrupted system call
sage:


HTH,
Ivan

[William Stein]

Hi,

As a test, can you guys try putting the file

http://wstein.org/home/wstein/tmp/fontList.cache

in your $HOME/.matplotlib directory?

William

[Aivo Jürgenson]

On Nov 19, 6:30 pm, William Stein <wst...@gmail.com> wrote:
> Hi,
>
> As a test, can you guys try putting the file
>
>          http://wstein.org/home/wstein/tmp/fontList.cache
>
> in your $HOME/.matplotlib directory?

The command line still fails

sage: plot(sin(x),0,4)
/Applications/sage/local/bin/sage-sage: line 203: 22376 Abort
trap sage-ipython "$@" -i

but in the browser notebook, the plot(sin(x),0,4) command doesn't give
any output and the console doesn't log anything bad either.

[Aivo Jürgenson]

Sorry, I was too hasty to report that. It seems that with cold sage startup, loading all the required libraries for plotting takes a bit of time and therefore the error message also took time to appear.

Actually, the plot(sin(x),0,4) command doesn't give any output on the browser notebook and in the console, the same error message is logged:

2009-11-28 13:06:36+0200 [HTTPChannel,11,127.0.0.1] got EOF subprocess must have crashed...

The Python crash report is also generated with the similar backtrace, so the fontList.cache didn't really help.

Aivo Jürgenson