[GRASS-dev] Raster history: why not reclass

[Markus Neteler]

On Sat, Sep 24, 2011 at 10:54 AM, Paolo Cavallini <cavallini at faunalia.it> wrote:
> Hi all.
> The history of rasters is one of the very nice features of grass. I
> noticed that in reclassified rasters, however, the rules are not saved,
> so the user does not have a clue of how the raster have been
> reclassified.

I agree.

> Is there any special reason for this? Any plan to implement it?

The reason will be the limited space in the history structure. The code
is essentially there:

[neteler at north r.reclass]$ grep hist *
reclass.c:    struct History hist;
reclass.c:    G_short_history(new_name, "reclass", &hist);
reclass.c:    strcpy(hist.datsrc_1, "Reclassified map based on:");
reclass.c:    sprintf(hist.datsrc_2, "  Map [%s] in mapset [%s]",
new.name, new.mapset);
reclass.c:    G_write_history(new_name, &hist);

http://grass.osgeo.org/programming6/history_8c.html#aadc6b56dba0ade914f5ed0c94600fa2a
http://grass.osgeo.org/programming6/structHistory.html
using
#define RECORD_LEN   80

http://grass.osgeo.org/programming6/dist_8i686-pc-linux-gnu_2include_2grass_2gis_8h.html#a244dc84fb432bea62b1716f74f1eb335

Not too many rules will fit in this 80 chars space.

Ideas
- improve this in GRASS 7
- store the reclass file in the metadata space in the mapset and display
  it with r.info if file present (should be backward compatible like this)

Markus