[GRASSLIST:6069] Re: How deep is a region?

[Hamish]

> cellhd.depths should be inited to 1, Cho followed gis/adj_cellhd.c
> where
> cellhd->depths =
>   (cellhd->top - cellhd->bottom + cellhd->tb_res/2.0) /
>   cellhd->tb_res;
> 
> I don't know why cellhd->tb_res/2.0, it is copy-paste of cols/rows.


Not fixed in CVS.  [lib/gis/adj_cellhd.c]


AFAIU, "depths" is the z analogue of "cols" and "rows" in the region
settings; ie number of grid cells on the z-axis. For 2D rasters we are
always one cell thick so this is going to be 1 for the majority of maps.

So cellhd->depths needs to be an interger, as does cellhd->rows and
cellhd->cols. 

but in the code:

cellhd->rows = (cellhd->north - cellhd->south + cellhd->ns_res/2.0) / cellhd->ns_res;
cellhd->cols = (cellhd->east - cellhd->west + cellhd->ew_res/2.0) / cellhd->ew_res;


.. it is leaving the cast to integer to the compiler. My C triva isn't
good enough to say for sure, but this is generally just an assumed
(int), right?

For clarity we should add cellhd->cols = (int) ((.. -+/2. ..)/ew_res);
if that is the intent? or should it be rounded instead of just stripping
digits to the right of the decimal point?

as for the /2.0 stuff, I don't know either. It has been like that for
rows & cols since the beginning of the CVS 5 years ago..


maybe Glynn [cc] understands.



Hamish