[GRASS5] d.legend updates

[H. Bowman]

I found myself using GIMP to place legends on raster images quite often, 
which didn't seem quite right, so I went ahead and wrote in the 
functionality that I wanted. And then I kept going..


New d.legend features include:

+  Legend text for floating point rasters is now much prettier.
      (%f -> %.2f, etc). Really tiny FP values convert to "2.4e-7" format.
+  at=x1,y1,x2,y2 implemented. Module is now scriptable / survives d.redraw
+  reimplemented -n flag: skip categories without labels. lines= takes on 
a new
      meaning when using this flag; means "up to category #". Its meaning 
becomes
      more blurry when you use lines=, thin=, and the -n flag together. 
This is
      a feature.
+  Floating point rasters now have the highest values at the top,
      lowest at the bottom.
+  Auto-scales text to fit in display window if bounds not manually set 
with
      at= or with mouse. Always auto-scales the "4 of 16 categories" text 
if
      needed.
+  You can now select the legend box with the mouse from any two corners
      and it will still work.
+  If you select a lower corner before a higher corner, the legend will
      reverse vertically. This also happens if the y values in at= are 
reversed.
      Works for horizontal legends as well.
+  fixed color-box off center within white frame for integer cats.
+  uses full legend box height to draw, even when using thin= or lines=.
+  better determines when to force a smoothed legend when using -n, lines=,
      &/or thin=.
+ smoothed legends with <5 categories now show <5 text labels.
+  Misc. code clean up & bug fixes, cleaned out some duplicate and
      unused code.



So, please try the new d.legend now in CVS, and let me know how badly you 
can break it, or what fine tuning you think it needs.

Hamish


p.s. - I've included a PNG of a legend I made with the PNG driver- it 
isn't quite right. Any ideas on if this would be a d.legend bug or a PNG 
driver bug? Both monitors were set to the same width and height (+/- 1 
pixel). Does PNG size need to match res from g.region for non-lossy 
output??

-------------- next part --------------
A non-text attachment was scrubbed...
Name: bad_png_driver.png
Type: image/x-png
Size: 4604 bytes
Desc: not available
Url : http://lists.osgeo.org/pipermail/grass-dev/attachments/20021129/17fac7c1/bad_png_driver.bin