[GRASS-dev] [bug #4522] (grass) r.series: error in row of input map is non-fatal giving wrong results

[Glynn Clements]

Request Tracker wrote:

> this bug's URL: http://intevation.de/rt/webrt?serial_num=4522
> -------------------------------------------------------------------------
> 
> Subject: r.series: error in row of input map is non-fatal giving wrong results
> 
> Hi,
> 
> 
> a problem with r.series:
> 
> One of my input maps has a corruption in it (disk flaw??), using r.series
> (method=average) I see this error:
> 
> WARNING: error reading compressed map [dbt_corrupt] in mapset
>          [sw2006f_icw], row 1465
> 
> but the process continues to completion and I get a resulting map which looks
> ok, but is subtly wrong.

> The read error should trigger a G_fatal_error().

Agreed. The question is: why does libgis leave this to individual
modules rather than doing it itself? Are there really modules which
might want to do something other than call G_fatal_error() in this
case?

FWIW, it isn't just r.series; the following all call G_get_raster_row
(and similar) without checking the return value:

display/d.rast.arrow
display/d.rast.edit
display/d.rast.num
display/d.rast
imagery/i.ifft
imagery/i.pca
lib/D
lib/ogsf
lib/rst/interp_float
paint/p.out.vrml
ps/ps.map
raster/r.bilinear
raster/r.carve
raster/r.contour
raster/r.flow
raster/r.grow2
raster/r.lake
raster/r.le/r.le.patch
raster/r.le/r.le.pixel
raster/r.le/r.le.setup
raster/r.le/r.le.trace
raster/r.neighbors
raster/r.param.scale
raster/r.random.cells
raster/r.random.surface
raster/r.recode
raster/r.series
raster/r.slope.aspect
raster/r.sum
raster/r.sun
raster/r.sunmask
raster/r.surf.area
raster/r.surf.contour
raster/r.texture
raster/r.to.vect
raster/r.volume
raster/r.water.outlet
raster/r.watershed/ram
raster/r.watershed/seg
raster/r.watershed/shed
raster/simwe/simlib
vector/v.vol.rst

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