This is now #12038, which has a patch which eliminates the segfault.
The problem was that you shouldn't be using coerce like that: the
first argument should be a parent, not the element class. That
element class requires two inputs: both the parent and the data to
make a new element out of. In this case, the __init__ method was only
getting one input (1+I) and thus raising a TypeError. There's now no
reference to your new ComplexNumber, so it's deallocated, but the
mpfrs are only initialized in the __init__ method, which was never
called. So when you try to clear them in __dealloc__, you get a
segfault.
David
On Mon, Nov 14, 2011 at 23:38, Justin C. Walker <jus...@mac.com> wrote:
>
> On Nov 14, 2011, at 18:04 , Dan Drake wrote:
>
>> This is not fun:
>>
>> sage: from sage.rings.complex_number import ComplexNumber as CN
>> sage: parent(1+I)
>> Symbolic Ring
>> sage: coerce(CN, 1+I)
>> BOOM
>>
>> (with 4.8.alpha1 and 4.7.2 in 64-bit Linux)
>>
>> It looks like mpfr is extremely unhappy about something.
>>
>> Also, if I just use the regular ComplexNumber in the global namespace
>> with the same command, I get "TypeError: unable to coerce to a
>> ComplexNumber: <type 'str'>". I understand that coercion may not be able
>> to handle SymbolicRing stuff to complex, but that's a strange message,
>> especially since ComplexNumber claims that it wants strings to convert.
>> But for now let's just deal with the segfault.
>
> I get segfaults for both cases ('CN' and 'ComplexNumber') all the way back to
> 2.11 (I've tried most of the 4.x releases, some of the 3.x's and only one of
> the 2's). I don't see the "TypeError" failure in any case.
>
> Justin
>
> --
> Justin C. Walker, Curmudgeon at Large
> Institute for the Absorption of Federal Funds
> --
> Democracy is two wolves and a lamb
> voting on what to have for lunch.
> Liberty is a well-armed lamb contesting
> the vote.
>
>
>
> --
> To post to this group, send an email to sage-devel@googlegroups.com
> To unsubscribe from this group, send an email to
> sage-devel+unsubscr...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/sage-devel
> URL: http://www.sagemath.org
>
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
