Re: Fwd: gfortran - pointers to components of derived type arrays

[Tobias Schlüter <tobias dot schlueter at physik dot uni-muenchen dot de>]

Paul Richard Thomas wrote:
> > PRs 29606, 30625 and 29396 are concerned with pointer assignments to
> > parts of a target array; either components or substrings.  As far as I
> > am aware, this is the last feature of the F95 standard that we do not
> > attempt to implement in gfortran.  Possible ways to do this, that I
> > know of, come in three flavours, in descending order of disruption to
> > the existing code:
> >
> > (i) To express the array stride in bytes rather than words;
>
> Ideally I would go for this. It allows pointers to funny array sections 
> to be
> represented with little or no runtime overhead.
>
> I expect that whichever option we choose there are going to be unexpected
> pitfalls. One advantage of changing the format completely is that you get
> much better test coverage because the same code is used for all pointer
> variables.

A minor remark: it should be possible to keep the old 
gfc_conv_descriptor_stride interface alive during the development (just 
have it include the necessary multiplication.  If that allows you to 
avoid a wholesale conversion of the code during development, it should 
make the whole thing much easier.

- Tobi