On Nov 11, 2007 9:06 PM, Robert Miller <[EMAIL PROTECTED]> wrote:
> I'm working on a new .pyx file, and everything has been working okay,
> but something strange is now happening, and I can't quite get to the
> bottom of it.
>
> Some side notes- I am running on OS X 10.5, a fresh install of
> sage-2.8.12. It failed building, I think it was scipy, but other than
> that it has been working ok.
Currently, Sage is not supposed to build completely on OS X 10.5
unless you install gfortran system-wide. (Don't ask me how to do
that.)
The only thing you loose as a result is scipy and maybe cvxopt (convex
optimization). Absolutely everything should have been built.
> So I have some new cython classes, and I've written some doctests.
> When I run through the doctests by hand, I can't reproduce what
> happens when I doctest: I run
> ./sage -t -verbose ......./relevant_file.pyx
> and I watch as every test passes, then it pauses for a while as it
> handles a bus error. Curious to see what is going on, I do the
> following:
>
> ./sage -t -gdb ......./relevant_file.pyx
>
> gdb says lots of things like
>
> warning: Could not find object file "/Volumes/HOME/sage/spkg/build/
> python-2.5.1.p8/src/Modules/python.o" - no debug information available
> for "./Modules/python.c".
>
> before it loads, but it loads,
That's I think completely normal on OS X.
> and I type "r" to run the tests. It
> complains more about symbols, such as
>
> Reading symbols for shared libraries warning: Could not find object
> file "/Volumes/HOME/sage/spkg/build/python-2.5.1.p8/src/build/
> temp.macosx-10.3-i386-2.5/Volumes/HOME/sage/spkg/build/python-2.5.1.p8/
> src/Modules/mathmodule.o" - no debug information available for "/
> Volumes/HOME/sage/spkg/build/python-2.5.1.p8/src/Modules/
> mathmodule.c".
>
> Then it runs the tests. The problem is
>
> Program received signal EXC_BAD_ACCESS, Could not access memory.
> Reason: KERN_PROTECTION_FAILURE at address: 0x00000015
> 0x000d1c50 in collect ()
>
> So I backtrace:
>
> (gdb) bt
> #0 0x000d1c50 in collect ()
> #1 0x000d276a in PyGC_Collect ()
> #2 0x000c18d0 in Py_Finalize ()
> #3 0x000c1376 in handle_system_exit ()
> #4 0x000c15e5 in PyErr_PrintEx ()
> #5 0x000c24ac in PyRun_SimpleFileExFlags ()
> #6 0x000d0d97 in Py_Main ()
> #7 0x00001fdc in _start ()
> #8 0x00001f09 in start ()
>
> I have reproduced what __dealloc__ does in another function, and have
> tried over and over to reproduce this error outside of doctest mode,
> to no avail. The fact that the garbage collector is accessing bad
> memory seems to indicate that it has already been freed or something,
> but this only only happens during doctests, or at least so it seems. I
> have run the doctests again and again under gdb by hand, and they
> never produce a problem, so I have it narrowed down to only every
> doctest in the file. The crappy thing is that I don't know how to have
> gdb tell me what the gc was trying to collect, and I can't reproduce
> the problem in an environment where I can narrow anything down.
>
> Given all these problems, I'm moving my project back to linux, for the
> stability of a happy install, and the extra memory tools. However,
> this situation is so confusing and bizarre I'd still like to figure
> out what is happening... Any help is appreciated
Questions:
(1) Do you or do you not actually have the same problem when
you build the same code under Linux? You don't say above.
(2) What happens when you do
grep "sage:" filename.pyx | sage
-- William
--~--~---------~--~----~------------~-------~--~----~
To post to this group, send email to sage-devel@googlegroups.com
To unsubscribe from this group, send email to [EMAIL PROTECTED]
For more options, visit this group at http://groups.google.com/group/sage-devel
URLs: http://sage.scipy.org/sage/ and http://modular.math.washington.edu/sage/
-~----------~----~----~----~------~----~------~--~---
