Type of the function argument(s) incorrect for GDALInvGeoTransform and GDALApplyGeoTransform

[yelnats]

The type of the function argument(s) is incorrect which is a pointer to a scaler, but requires an array”

< $ffi->attach('GDALInvGeoTransform' => [qw/double[] double[]/] => 'int');
< $ffi->attach('GDALApplyGeoTransform' => [qw/double[] double double double* double*/] => 'void');

$ffi->attach('GDALInvGeoTransform' => [qw/double* double*/] => 'int');
$ffi->attach('GDALApplyGeoTransform' => [qw/double* double double double* double*/] => 'void');

[fpl]

The same issue appears even for CT* functions, where prototypes should be as follows:

 $ffi->attach('OCTTransform' => [qw/opaque int double[] double[] double[]/] => 'int');
 $ffi->attach('OCTTransformEx' => [qw/opaque int double[] double[] double[] int[]/] => 'int');
 $ffi->attach('OCTTransform4D' => [qw/opaque int double[] double[] double[] double[] int[]/] => 'int');
 $ffi->attach('OCTTransform4DWithErrorCodes' => [qw/opaque int double[] double[] double[] double[] int[]/] => 'int');
 $ffi->attach('OCTTransformBounds' => [qw/opaque double double double double double[] double[] double[] double[] int/] => 'int');

This isn't very pleasant because it renders the whole coordinate transformation unusable.

[shawnlaffan]

A PR would help clarify the changes needed.

The underlying issue might be in the header processing script, though, so it is perhaps best fixed there. That might best be done by Ari.
https://github.com/ajolma/Geo-GDAL-FFI/blob/master/build-tools/parse_h.pl

[fpl]

About parse_h.pl apparently it has a little issue with the parsing, because the double* should not be overridden for all args as in its sub parse_type with %use_array, else the fix would be quite trivial.
EDIT: I mean the whole logic of parse_type should change using some trick such as using an additional argument counter.