On 30 June 2010 08:29, Robert Bradshaw <[email protected]> wrote:
> On Jun 28, 2010, at 10:20 AM, luisfe wrote:
>
>> Hi,
>>
>> I have found an unhandled SIGFPE in number_field_element_quadratic as
>> explained in ticket http://trac.sagemath.org/sage_trac/ticket/9357
>>
>> Basically, sage does not check if a quadratic algebraic number is zero
>> when trying to invert it.
>>
>> I added a trivial patch that checks if the zero element is being
>> inverted to rise a ZeroDivisionError. However, the error message
>> before the crash suggests that the compiled code is not properly
>> wrapped with _sig_on _sig_off.
>>
>> In this example even if one adds this wrapper the zero check is
>> advisable, since _sing_on would rise a RuntimeError instead a
>> ZeroDivisionError, so it seems that the zero check is enough in this
>> case. On the other hand would the _sig_on stuff made the code more
>> robust? What is the way to proceed in this cases?
>
> You did the right thing. The _sig_on and _sig_off are more about being able
> to trap signals such as interrupts (so that control-C works).
>
> - Robert

Is there any quick way of finding code which might not be wrapped
properly with sig_on and sig_off? I think the seg fault I get at
startup indicates that as a possible problem. (Not sure. Forget error
message and can't check now. Batteries just about to die in laptop)

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

Reply via email to