PETScMatrix doesn't work properly from Python

[Garth Wells]

Some Python demos (e.g. la/eigenvalue/python) are failing because of a problem with the use of PETScMatrix (Matrix seems to be fine).

The error during assembly is weird:

python: /home/garth/code/fenics/dolfin.d/dolfin-all/dolfin/la/PETScMatrix.cpp:271: virtual void dolfin::PETScMatrix::setrow(dolfin::uint, ....

PETScMatrix::setrow shouldn't be called. Is there a problem with the SWIG typemaps (that is't trying to copy data)?

[Garth Wells (garth-wells)]

Could it be an issue that PETScMatrix is a subclass of PETScBaseMatrix as well as GenericMatrix? (in terms of the SWIG wrappers)

[Garth Wells (garth-wells)]

Fixed in http://bazaar.launchpad.net/~dolfin-core/dolfin/main/revision/5562

[Johan Hake (johan-hake)]

Great you manage to solve it.

It worked fine here so I did not know where to start.

Johan

On Saturday January 22 2011 08:15:52 Garth Wells wrote:
> Public bug reported:
>
> Some Python demos (e.g. la/eigenvalue/python) are failing because of a
> problem with the use of PETScMatrix (Matrix seems to be fine).
>
> The error during assembly is weird:
>
> python: /home/garth/code/fenics/dolfin.d/dolfin-
> all/dolfin/la/PETScMatrix.cpp:271: virtual void
> dolfin::PETScMatrix::setrow(dolfin::uint, ....
>
> PETScMatrix::setrow shouldn't be called. Is there a problem with the
> SWIG typemaps (that is't trying to copy data)?
>
> ** Affects: dolfin
> Importance: Critical
> Assignee: Johan Hake (johan-hake)
> Status: Confirmed
>
> ** Changed in: dolfin
> Importance: Undecided => Critical
>
> ** Changed in: dolfin
> Status: New => Confirmed
>
> ** Changed in: dolfin
> Assignee: (unassigned) => Johan Hake (johan-hake)

[Garth Wells (garth-wells)]

On 22/01/11 20:16, Johan Hake wrote:
> Great you manage to solve it.
>
> It worked fine here so I did not know where to start.
>

Do you have Swig version 2?

Garth

> Johan
>
> On Saturday January 22 2011 08:15:52 Garth Wells wrote:
>> Public bug reported:
>>
>> Some Python demos (e.g. la/eigenvalue/python) are failing because of a
>> problem with the use of PETScMatrix (Matrix seems to be fine).
>>
>> The error during assembly is weird:
>>
>> python: /home/garth/code/fenics/dolfin.d/dolfin-
>> all/dolfin/la/PETScMatrix.cpp:271: virtual void
>> dolfin::PETScMatrix::setrow(dolfin::uint, ....
>>
>> PETScMatrix::setrow shouldn't be called. Is there a problem with the
>> SWIG typemaps (that is't trying to copy data)?
>>
>> ** Affects: dolfin
>> Importance: Critical
>> Assignee: Johan Hake (johan-hake)
>> Status: Confirmed
>>
>> ** Changed in: dolfin
>> Importance: Undecided => Critical
>>
>> ** Changed in: dolfin
>> Status: New => Confirmed
>>
>> ** Changed in: dolfin
>> Assignee: (unassigned) => Johan Hake (johan-hake)
>