#1804: r.clump NULL handling
-------------------------+--------------------------------------------------
Reporter: marisn | Owner: grass-dev@…
Type: defect | Status: new
Priority: normal | Milestone: 7.0.0
Component: Raster | Version: svn-trunk
Keywords: | Platform: Unspecified
Cpu: Unspecified |
-------------------------+--------------------------------------------------
Current r.clump implementation uses 0 as NULL indicator and NULL cells as
any other values.
Setting 0 and NULL to NULL can be done with following code. Unfortunately
today my pointer-fu is too weak (around line 229) to change code to treat
0 as valid value and leave NULL as-is.
Clumping 0 areas as any other areas would be sensible thing. The open
question remains - should NULL areas also be clumped or just skipped and
left as-is (NULL)?
{{{
Index: raster/r.clump/clump.c
--- raster/r.clump/clump.c (revision 53904)
+++ raster/r.clump/clump.c (working copy)
@@ -105,7 +105,7 @@
for (col = 1; col <= ncols; col++) {
LEFT = X;
X = cur_in[col];
- if (X == 0) { /* don't clump zero data */
+ if (X == 0 || Rast_is_c_null_value(&X)) { /* don't clump
zero/NULL data */
cur_clump[col] = 0;
continue;
}
}}}
Fixes have been made in r59076, r59082, r59122, r59128, r59131, r59136 but
I don't know if
r.clump2 in Addons is still better. The first fixes "do not clump NULL
data".
Replying to [comment:4 neteler]:
> Fixes have been made in r59076, r59082, r59122, r59128, r59131, r59136
but I don't know if
> r.clump2 in Addons is still better. The first fixes "do not clump NULL
data".
r.clump2 did not provide all the functionality of r.clump and was slower
than the new r.clump; removed from addons.
Maybe replace r.clump in