ffc team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

We're getting closer. How did you check that the quadrature code is
correct? I get a lot of error messages in error.log still. Or are
those from evaluate_basis_* functions and tensor representation?

--
Anders


On Tue, Jan 26, 2010 at 10:12:16PM -0000, noreply@xxxxxxxxxxxxx wrote:
> ------------------------------------------------------------
> revno: 1555
> committer: Kristian B. Ølgaard <k.b.oelgaard@xxxxxxxxx>
> branch nick: dev
> timestamp: Tue 2010-01-26 22:42:59 +0100
> message:
>   Modify format['transform'], quadrature code compiles and is correct for all forms except QuadratureElement and ElementRestriction.
> modified:
>   ffc/cpp.py
>
>

> === modified file 'ffc/cpp.py'
> --- ffc/cpp.py	2010-01-26 15:11:26 +0000
> +++ ffc/cpp.py	2010-01-26 21:42:59 +0000
> @@ -253,6 +253,11 @@
>  def _transform(type, j, k, r):
>      # FIXME: j, k might need to be swapped for J or JINV
>      map_name = {"J": "J", "JINV": "K"}[type] + {None: "", "+": "0", "-": "1"}[r]
> +    # FIXME: KBO: In the old format["transform"] we swapped j and k, in the old
> +    # format["transform_ufl"] we didn't. format["transform_ufl"] was only used
> +    # in the tensorgenerator so it might not be needed anymore?
> +    if type == "J":
> +        return (map_name + "_%d%d") % (k, j)
>      return (map_name + "_%d%d") % (j, k)
>
>  def _generate_switch(variable, cases, default=None):
>

Attachment: signature.asc
Description: Digital signature