[GRASS-dev] Error reading null row 2725 for <MASK>

[Markus Neteler]

On Thu, Aug 28, 2014 at 7:34 AM, Paulo van Breugel
<p.vanbreugel at gmail.com> wrote:
> On Wed, Aug 27, 2014 at 8:20 PM, Markus Neteler <neteler at osgeo.org> wrote:
>> On Wed, Aug 27, 2014 at 3:25 PM, Paulo van Breugel
>> > I forgot to mention that I am using GRASS 7.1 (rev 61757, had it before
>> > with rev 61413, but I think I had a similar problem about a year ago).
>> >
>> > On Wed, Aug 27, 2014 at 10:13 AM, Paulo van Breugel
>> >> When running several r.mapcalc in a loop (via R) the function stops
>> >> with an error:
>> >>
>> >> ERROR: Error reading null row 2725 for <MASK>
>>
>> The error comes from read_null_bits() in
>> lib/raster/get_row.c
>>
>> around
>> http://trac.osgeo.org/grass/browser/grass/trunk/lib/raster/get_row.c#L827
>>
>> Maybe some off_t definition is missing?
>
> I don't know what that means; indication of something being wrong with the
> data, or in the code?

Question: did you compile with --with-pthread?
AFAIK it causes troubles.


Markus

>> >> The row number varies. This seems to happen at random, and when I
>> >> repeat
>> >> the same r.mapcalc calculation that stopped with the error before,
>> >> things run without problem.
>>
>> Please tell us also
>> - how many raster cells are involved
>> - how many files are opened in the loop?
>
> Last time the error occured in the r.mapcalc below:
>
>   The command:
> r.mapcalc --overwrite expression="OUTPUT = if(MESSC==0, 100000* 100.0 *
> (MESSB-12300000)/(382900000-12300000), if(MESSC<=5000000, 2*MESSC,
> if(MESSC<=10000000, 2*(10000000-MESSC),100000 * 100.0 * (382900000-
> MESSB)/(382900000-12300000))))"
> produced an error (1) during execution:
> ERROR: Error reading null row 2725 for <MASK>
>
> These layers have 6515140 raster cells, with 3366825 total null cells. The
> command is part of a sequence of commands that is run in a loop, but these
> are run in sequence so at the time of the error I guess two layers were
> open. As I mentioned, running the command again and it will normally run
> without problems with the given input layers, and it may go wrong with other
> input layers again, without any discernible logic to me.
>
>>
>> >> I tried after removing the mask, and this time no problem so it may
>> >> have
>> >> to do something with the MASK? (as these error messages only occur once
>> >> in a
>> >> while, I can't be completely sure it only happens when a MASK is
>> >> defined).
>> >>
>> >> I ran it through R, using the execGRASS() function of the spgrass6
>> >> package. I ran the same, but using the system call instead of
>> >> execGRASS.
>> >> Still got the same error messages. I therefore guess this has nothing
>> >> to do
>> >> with R?
>>
>> Could you "simulate" the loops in a GRASS script outside of R to see
>> if it the same?
>
>
> I'll have to figure out how to do that. Perhaps running the r.mapcalc above
> several times will do the trick.
>>
>>
>> Markus
>
>