 |
 |
Tim, > >> ARR_xTOx( USEBAD, NDIMS, DIMS, LBND, UBND, INARR, OUTARR, IERR, NERR, > >> STATUS) > >> > >> (with USEBAD, IERR and NERR simply there to match the prm interface for > >> VEC_). Then if LBND is all ones and UBND == DIMS then this is a vec_ > >> copy, else it's a manual copy of a subarray (it's actually a VEC_ > >> copy if the first NDIMS-1 dims match DIMS). > > > > Sounds good to me. The only other suggestion may be to replace the DIMS > > argument by a LBND_IN/UBND_IN pair since otherwise the LBND(_OUT)/UBND(_OUT) > > values would need to be supplied in GRID indicies rather than PIXEL > > indices. I know its not a big deal to convert backwards and forwards > > between GRID and PIXEL, but it is at least an annoyance and introduces > > scope for bugs. > > > > But then don't I have to address a negative index in a fortran array? I was imagining that ARR_xTOx would calculate the DIMS corresponding to the supplied LBND_IN/UBND_IN, then convert the supplied LBND(_OUT)/UBND(_OUT) from PIXEL to GRID indices, then do the copy using the calculated DIMS and GRID bounds. I was only saying that it would be better for these PIXEL<>GRID conversions to be done inside ARR_xTOx rather than outside. > OUTARR(-1, -20) = INARR(-1, -20) > > looks a bit dangerous to me.... [I realise that I have to trust the caller > tha NDIMS is actually within the declared bounds of the array...] But presumably you would be accessing both arrays as 1D vectors within ARR_xTOx and calculating the offsets yourself? > I was intending LBND/UBND to be the array indices (GRID coordinates) of > both the input and output, not a mapping from one region of the input to a > different region in the output. Feel free to convince me of complexity > though. No complexity - no Mappings. I'm just saying that if you have (say) a 2D NDF with bounds (-100:50,120:200) and you want to copy the section (-50:10,140:160) to a new NDF, then an application writer would like to define the shape of the input array using the bounds LBND_IN=[-100,120] UBND_IN=[50,200] rather than having to calculate the dimensions DIMS=[221,151], and would also like to supply LBND(_OUT)=[-50,140] and UBND(_OUT)=[10,160] to define the section to be copied, rather than having to convert these bounds from PIXEL indices to GRID indices. David