MINDISTANCE not considering label size on lines

[adube]

Distance between labels do not consider the label size. If you have a long label on a line with a MINDISTANCE of let's say 80, we don't have 80 pixels between labels. Instead, we have 80 pixels between the position of the labels.

Since the new REPEATDISTANCE do consider the label size, it should be the same thing with MINDISTANCE.

I'll come up with a sample data test that demonstrates the problem.

[adube]

MINDISTANCE 80, should be 80px between Dudley Ave, currently not

[adube]

A sample dataset showing the bug.

[aboudreault]

Since the label size is calculated before the calls of msTestLabelCacheCollisions(), we could simply add it as a parameter to the msTestLabelCacheCollisions() function. Then, we verify that the distance between the 2 points is <= than the mindistance value + half the size of the label (ie. if the label position if CC). What do you think?

[dmorissette]

Sounds like an acceptable fix. Adding Steve to CC to get his opinion.

[sdlime]

Not perfect but better than what's there and a heckuva lot faster than checking the true distance between a cached label's polygon. Have you tried it?

Steve

[aboudreault]

Not yet tried. Will test the proposed method in the next days.

[aboudreault]

Seems to work well, I've simply added the label_size to the mindistance value when checking the point to point distance. That way, I ensure that there is really the mindistance value between the two labels

[aboudreault]

Fixed and committed in r9198.