wrong dash pattern

[zanollim]

I'm not able to reproduce this pattern [mapserver-6.0.0-beta3]:

==== == ==== == ==== ==

here the map code

STYLE
  COLOR 0 0 0
  LINECAP square
  PATTERN 4 2 2 2 END
END

Is this a bug? Thanks, michele

[tbonfort]

I'm guessing that the square linecap is eating into the blank space left between the dashes. Try with linecap butt, or increasing the spacing, eg:

4 3 2 3

[zanollim]

Thanks tbonfort,

it seems that 4 2 1 2 is the best solution for me.
But in the map the value 2 [num off] is shorter than value 1 [num on]
(see the screenshot attached) I think this is a wrong behavior

michele

[tbonfort]

with LINECAP square I don't think it is the wrong behavior

[zanollim]

but my screenshot refers exactly to this code

STYLE
  WIDTH 3
  OUTLINECOLOR 160 0 160
  LINECAP butt
  PATTERN 4 2 1 2 END
END

[zanollim]

pattern 4 3 1 3 (crazy behavior)

[tbonfort]

Ok,

there might be some glitches in how different renderers interpret the line caps. Given that there's a workaround (by augmenting the white space), and that the implementation of a fix would involve getting into the code of each renderer (agg,cairo,gd), I don't think this this will get investigated more.

[zanollim]

As you can see in the pattern_4_3_1_3 attachment, augmenting the white space returns a copletely different and wrong behavior.

[tbonfort]

Note that the pattern, starting from beta3, is not scaled by the line width (as was the case beforehand). In that regard, it is usually not a good idea to have any value in your pattern that is smaller than the actual width of your line.

so, using width 3, do you get what you expect with

STYLE
  WIDTH 3
  OUTLINECOLOR 160 0 160
  LINECAP butt
  PATTERN 12 6 6 6 END
END

?

[zanollim]

sorry tbonfort, but your style code (see pattern_12_6_6_6 attachment) is too far from the best solution for me (see the dash_dot attachment)

[tbonfort]

thank you for pushing on this ;)

I looked in the code and there was a bug in the mapserver agg code. I'll commit the fix asap and let you try with trunk.

[tbonfort]

committed in r11364.

please report back.

[zanollim]

thank you for the very immediate support!!! I will try soon

[zanollim]

My tests with mapserver-6.0.0-beta4 are OK! For me this ticket can be closed