Andy Porter

Results 146 issues of Andy Porter

Mixed precision kernels have multiple subroutines defined in an interface block. This PR extends get_kernel_schedule to work with such kernels.

in progress
LFRic

The latest fparser has additional symbol table functionality and support for some additional Fortran2008 features (particularly the `mold` argument to `allocate`).

in progress
GOcean
LFRic
NEMO

In #1366 I'm adding support for the Allocate intrinsic and as part of that I want to support the 'mold' parameter. That's an F2008 feature and the version of fparser...

enhancement
in progress
GOcean
LFRic
NEMO
NEMOVAR

Currently we default to expecting the Fortran source containing a particular module to have the same name as that module. This is clearly a bit of a limitation. I guess...

enhancement
PSyIR

There are a number of places where we use a writer when constructing error messages. At the moment the writer to use normally defaults to `FortranWriter` but can b overridden...

question

As noted here https://github.com/stfc/PSyclone/pull/1109#discussion_r583761364, it would be useful if `process_nemo.py` (in examples/nemo/scripts) was generalised so as to support batch processing of any application. The only NEMO-specific code in that script...

enhancement

Simon is finding lots of bugs. For the following code: PROGRAM x INTEGER :: y=MAX(1,2,3) DO y=1,2 END DO END PROGRAM x `y` ends up being of `UnknownFortranType` (because of...

bug
in progress
NEMO
PSyIR

#1782 is bringing the first cut of the PSyAD test-harness generation for LFRic onto master. However, it does not yet support kernels with operator arguments.

adjoint

In #1782 I've had to enable line-length limiting to get generated code that is accepted by the LFRic build system. Really this should be a command-line flag.

enhancement
adjoint