[GRASS-user] displaying many thematic maps from one vector dataset

[Vishal Mehta]

Hi Moritz,
Regarding legends with d.graph, these are the main issues where I found I
needed to do too much of trial and error, some of which I think could
probably be avoided/improved in a future version...

1. the legend ends up with some strange additions beyond just the range of
data: like what are those numbers after each range
e.g. in the example below where did | 0 and | 5 come from? It would be nice
to not have this automatically displayed.
60 - 80 | 0
80-100 | 5

2. there could be a flag on d.graph, which allows the output to have a "<
minbreak" and "> maxbreak" option, so that for cases like the one we are
discussing, the legend does not keep changing based on different data
ranges.

3. size and position
Here, I wanted to get a legend file separately, and save a legend png. To do
this I had to play with the size and other parameters extensively to make
the symbols and associated text match up. It would be great to be able to do
the following with some flags, for the case where the legend alone is
displayed:
(i) set the overall size of the display within which the legend will be
displayed
(ii) set the position of the legend within the display with options like
lowerleft, upperright, center, etc
(iii) set the size of the legend as a fraction/percentage of the display.
(iv) make the background transparent (or not)
such that the above settings 'automatically' set up the symbol and
corresponding text all properly lined up..

I am aware that if I was just better at this I could do it quickly
enough..but the above functionality would make it easier for
less-than-expert people like myself, who try to use open-source as much as
possible despite the handicap of having been groomed on Windows...

I will write back when i've had a chance to look up the segfaults on my
laptop..

And oh, the mismatch of color between display and legend was entirely my
fault - i was displaying with one set of colors but had tweaked the legend
file separately with a different set of colors..

Thanks for all your help,
Vishal

On Wed, Apr 8, 2009 at 4:58 AM, Moritz Lennert <mlennert at club.worldonline.be
> wrote:

> On 08/04/09 02:23, Vishal Mehta wrote:
>
>> Regarding the segmentation faults: I switched to a different machine
>> to do all this latter work. I havent tried the new scripts on the
>> original machine (my laptop). Both are running 6.4svn.
>>
>
> Weird. I can confirm segfaults in 6.5svn (I suppose that this is what
> you mean by 6.4svn, i.e. the current 6.x development branch), but only
> when using d.mon, not when using direct rendering
> (GRASS_RENDER_IMMEDIATE=TRUE). Since I wrote this code on the base of
> the grass7 tree, and then ported it back to 6.x, it is quite probable
> that I forgot some adaptations needed for d.mons... (CC'ing to Glynn for
> any pointer on where I should look).
> Could you try with direct rendering on the machine where it segfaulted
> with d.mons ?
>
>  Regarding legends, I found that I had to play around way too much to
>> get the legend right.
>>
>
> Could you tell me which were the major parts that needed adjustment from
> you ?
>
> As you want to use the same legend for all maps, it is obviously easier
> to create one legend file by hand and to call d.graph on that in each cycle
> of the loop.
>
> However, I tweaked the source code (grass7 for now) just very slightly
> (changing text size and taking into account the case where the min value of
> the data is above the lowest break (or the max value below the highest break
> - see 2010-2030 max values in the example) and then ran the following script
> (using direct rendering as monitors don't exist anymore):
>
> export GRASS_FONT=Vera
> export GRASS_PNG_READ=TRUE
>
> yr=2010
> while [ $yr -lt 2105 ] ;
>  do
>    echo "yr is $yr "
>    export GRASS_PNGFILE=pc$yr.png
>    d.thematic.area map=CA column=pc$yr breaks=20,40,60\
>                colors=cyan,yellow,red,blue leg=legfile
>    d.text -b text="$yr Population" at=50,90 size=4
>    d.graph in=legfile
>    if [ $yr -eq 2090 ]
>       then yr=$((yr+10))
>       else yr=$((yr+5))
>    fi
>  done
>
>
> You can see the resulting animated gif here:
> http://geog-pc40.ulb.ac.be/popCA.gif
>
> I don't find the legends too bad, but obviously it is difficult to create a
> one-size-fits-all solution for legends. I spent some time trying to find the
> right algorithm, but it is still very far from perfect. One thing I should
> probably add is a parameter for the placement of the legend in screen
> percents. Another is a detection mechanism to see how many significant
> decimals to keep (There should be some examples, notably in Hamish'es work,
> I just need to look at those...). I don't know, however, whether the d.graph
> solution is the best, so feedback is very welcome on that.
>
>
>  And even then, there's a yellow color that is not on the legend file,
>> that shows up in the display anyway.
>>
>
> Could you show an example of that ?
>
> Thank you for your feedback so far. We really have to identify the cause of
> those segfaults...
>
> Moritz
>



-- 
Vishal K. Mehta, PhD
Scientist
Stockholm Environment Institute - US
133 D St Suite F
Davis CA 95616
www.sei-us.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/grass-user/attachments/20090408/16a715d5/attachment.html