r.clump: inadequate color ramp

[pcav]

r.clump applies a color table to results, but if we have many clumps (>900 in my case), higher clumps do not get a color, and appear (wrongly) missing.

[hamish]

after closing the map r.clump does:

    G_read_range(OUTPUT, G_mapset(), &range);
    G_get_range_min_max(&range, &min, &max);
    G_make_random_colors(&colr, min, max);
    G_write_colors(OUTPUT, G_mapset(), &colr);

so maybe G_make_random_colors() is to blame?

it falling over at ~900 clumps is suspiciously close to this calculation in lib/gis/color_rand.c:

#define MAX_COLORS 1024
#define DEVIATION 128
...
count = MAX_COLORS - DEVIATION + rand() % DEVIATION;

as a basis for testing, could someone who knows about the module add an example using one of the standard datasets to the help page?

see also r.watershed's color creation mechanism.

thanks, Hamish