[GRASS-dev] r.univar: allow multiple rasters to be processed

[Glynn Clements]

Ivan Shmakov wrote:

>  > Is there some fundamental reason why r.univar has separate cases for
>  > CELL/FCELL/DCELL types, rather than just working in DCELL throughout?
> 
>  > Most of the time, there's no value to having separate CELL/FCELL
>  > cases. There's hardly ever a reason to have a separate FCELL case,
>  > and a separate CELL case is only really useful if you specifically
>  > need integer-specific operations (e.g. integer division, bitwise
>  > operations etc).
> 
>  > I'm aware that r.univar was already like this, but if there isn't a
>  > valid reason, this would be a good time to simplify it.
> 
> 	I don't know for sure, but apparently the distinction is only
> 	effective when gathering extended statistics (percentiles,
> 	median.)
> 
> 	Since the extended statistics code temporarily stores all the
> 	values from the raster(s) being processed, using native types
> 	may lead to reduced memory usage and processing time.

I doubt that it will make much difference regarding overall processing
time.

So far as memory usage is concerned: if we think that people might
want to compute quantiles on data sets which are large enough that we
need to worry about memory consumption, we should probably be looking
for a more efficient algorithm. Sorting the entire data set then
pulling out quantiles is less than ideal if you're dealing with that
much data.

-- 
Glynn Clements <glynn at gclements.plus.com>