[GRASS-dev] 6.4.0 -ending story

Markus Metz markus.metz.giswork at googlemail.com
Wed Jan 6 04:13:56 EST 2010


Hamish wrote:
> Michael:
>   
>>> A big blocker was fixed yesterday - v.delauany broken on Mac and
>>> Windows.
>>>       
>
> (everyone needs to test the heck out of it now of course)
>   
Suggested testing in spearfish:
start with the reduced archsites vector provided by Michael, 11 points:
https://trac.osgeo.org/grass/attachment/ticket/660/archsites11.zip

next the full archsites map, 25 points, then bugsites, 90 points

In the North Carolina dataset, you could test with elev_lid792_bepts, 
118,716 points

If everything is fine so far, then generate 3 million random points and 
test that. Be aware that the output vector will be very large if area 
output is requested, rather use v.delaunay -l to get only lines.

In all test, v.delaunay should be run with --verbose in order to be able 
to follow the steps. The Delaunay triangulation should always finish 
within seconds, for millions of points < 1 min, whereas topology 
building can take some time.
>   
>> * 759 r.patch non-functional in WinGRASS 6.4 svn on Vista
>>   -> awaiting guru comment
>>     
> Michael:
>   
>>> This is due to security routines on *some* Vista systems.
>>>       
>
> I'm not totally convinced of our analysis of that (see my last comments
> in the bug report). but if the work-around does the trick that's good
> enough for the release notes.
>   
The currently best guess is that this is the Vista UAC problem. Provide 
a comment and tips for Vista UAC in a prominent place for the release?
>  
>   
>> * 843 v.digit broken on new WinGrass release
>>   -> contains a suggestion
>>     
>
> Maris and Glynn are working to resolve this. Seems plausible that it 
> will be backported and ready for final testing by next week.
>   
One of the vector digitizers *must* work for the final release. The 
wxdigitizer is a fairly new feature and IMHO should not be a blocker. 
That said, I personally use only the wxdigitizer and no longer v.digit.
>   
>> * 384 wxGUI: vdigit crashes on a big map
>>   -> who can confirm?
>>     
>
>   
Testing here with output of v.delaunay for elev_lid792_bepts (237,392 
areas), it is not exactly crashing, it is forever busy with something. 
Building topology with v.build takes here about 5 minutes, the 
wxdigitizer needed more than 30 minutes just to display the map. When 
exiting the wxdigitizer, topology was built by the wxdigitizer, but much 
slower, factor >10, so I killed the wxdigitizer. Something in the 
wxdigitizer causes a serious slowdown that can appear like a crash 
because nothing is happening. That was on Linux 64 bit.
>   
>> * 461 v.to.db crashes on a shapefile connected with
>> v.external
>>  -> still valid?
>>     
>
>   
I think so because attribute queries are not possible on OGR vectors 
connected with v.external (most OGR vector layers don't have a 
grass-like key column). I think Martin L fixed that in grass7.
>   
>> I personally vote for going ahead and publishing this week RC6
>>     
>
> My vote is to wait for the v.digit stuff to get sorted out, and as
> soon as it is cut the RC.
>   
The wx location wizard is still not working properly, I still can't 
specify UTM zone. If the wxGUI is supposed to be stable, that should IMO 
get fixed before the final release. This is work in progress, ticket #842.

my 2c,

Markus M


More information about the grass-dev mailing list