this bug's URL: http://intevation.de/rt/webrt?serial_num=863
-------------------------------------------------------------------------
Subject: 2 bugs in GIS lib: reclass maps
grass downloaded at: CVS Server, Germany
Hi again,
I have found two bugs in the GIS library: The reclass-test function is
not working properly when a raster and and vector map with identical
name exist (which is usually no problem). And you can't delete the base
map of a reclass map, if you first delete the reclass map (was reported in
RT as well). Please have a look at the situation:
My maps are:
g.list rast
----------------------------------------------
raster files available in mapset user1:
muell
-> which is a common map (has been base to another, already deleted, reclass
map).
r.info muell
| Title: clump of vegcover in PERMANENT ( vegcover.clump )
|----------------------------------------------------------------
|
| Type of Map: raster Number of Categories: 3230
| Data Type: CELL
And there is:
g.list vect
----------------------------------------------
vector files available in mapset user1:
muell
Now it happens:
g.remove vect=muell
[muell@user1] is a base map. Remove reclassed map first:
vegcover.1000@user1
Oops: I used "vect="!
strace g.remove vect=muell
[...]
close(3) = 0
munmap(0x40017000, 4096) = 0
brk(0x8087000) = 0x8087000
open("/home/neteler/grassdata/spearfish/user1/cell_misc/muell/reclassed_to", O_RDONLY) = 3
fstat64(3, {st_mode=S_IFREG|0644, st_size=20, ...}) = 0
old_mmap(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0
[...]
huh - this looks really nasty.
And, to make it complete:
g.remove rast=muell
[muell@user1] is a base map. Remove reclassed map first:
vegcover.1000@user1
Also not working as vegcover.1000@user1 is already deleted (see above).
Reason is the left-over cell_misc/muell/reclassed_to which wasn't cleaned
up when I deleted vegcover.1000@user1 (here the entry must be deleted, not the
file I guess as a map may be base for several reclass files).
So far my report. Again looking for a volunteer here (too complex for me).
BTW: I have resolved the reported g.mremove/mapset problem. It is updated
in CVS.
Best regards
Markus
-------------------------------------------- Managed by Request Tracker