horizontal lines with larger dataset

[richlv]

normal view, not wireframe one.

sometimes horizontal lines appear between dataset points. usually seems to be triggered by having more nodes in a single way.

a couple of screenshots attached - one showing the glitch with "normal" dataset, and one with gpx converted to data layer (as this allows for a very easy reproduction of the error).

changing active layer or layer ordering does not impact the glitch, changing zoom and panning around does.

Repository Root: http://josm.openstreetmap.de/svn
Build-Date: 2011-04-14 01:31:50
Last Changed Author: bastiK
Revision: 4021
Repository UUID: 0c6e7542-c601-0410-84e7-c038aed88b3b
URL: http://josm.openstreetmap.de/svn/trunk
Last Changed Date: 2011-04-13 22:43:46 +0200 (Wed, 13 Apr 2011)
Last Changed Rev: 4021

Identification: JOSM/1.5 (4021 en)
Memory Usage: 133 MB / 682 MB (29 MB allocated, but free)
Java version: 1.6.0_20, Sun Microsystems Inc., OpenJDK Server VM
Operating system: Linux
Dataset consistency test: No problems found

Plugin: PicLayer (25219)
Plugin: SimplifyArea (25560)
Plugin: livegps (25833)
Plugin: osmarender (25371)
Plugin: routing (25843)

[richlv]

error with "normal" dataset

[richlv]

error with converted gpx

[richlv]

there are multiple older issues about similar glitches, but they are different (and have been already fixed).
also, doesn't seem to be reproducible with tested-3966 (but maybe that's just "bad" luck)

[stoecker]

Java version: 1.6.0_20, Sun Microsystems Inc., OpenJDK Server VM

Is this now OpenJDK or Sun? OpenJDK has more problems with this type of issues. Usually it is fixable by clipping the display (in this case the bug is somewhere in gpx displaying again). Java has troubles, when displayed stuff is a lot outside of the visible area.

[richlv]

that's now it appears in josm report :)
package name is "java-1_6_0-openjdk", so maybe it adds "sun" part to deal with some apps hardcoding jvm name... just guessing here.

note that it's not with gpx displaying but with data layer displaying.

[skyper]

Replying to richlv:

there are multiple older issues about similar glitches, but they are different (and have been already fixed).
also, doesn't seem to be reproducible with tested-3966 (but maybe that's just "bad" luck)

#6050 is still open.

[bastiK]

In [4042/josm]:

fixed #6218 - horizontal lines with larger dataset (openjdk does not like NaN in Path2D)