symbol with (x -99) offset causes wrong rendering

[diletant]

inside a map file, if you assign an offset (type x -99) on several symbols from a same class, some non-existing pattern are graphically rendered. For instance, it seems that linear segments are prolonged further than their real end.

This issue doesn't appear if only one offset (x -99) symbol is drawn for a class item.

[diletant]

And the wrongly rendered's lines are changing with scale (bbox related)

[diletant]

Replying to diletant:

inside a map file, if you assign an offset (type x -99) on a symbol, some non-existing patterns are graphically rendered. For instance, it seems that linear segments are prolonged further than their real end.

[tbonfort]

Replying to diletant:

Replying to diletant:

inside a map file, if you assign an offset (type x -99) on a symbol, some non-existing patterns are graphically rendered. For instance, it seems that linear segments are prolonged further than their real end.

hi, could you come up with a limited test case (data+mapfile) that exhibits this, as I can't reproduce this with what I have. Also, does this happen when using the GD renderer ?

[diletant]

Replying to tbonfort:

Replying to diletant:

Replying to diletant:

inside a map file, if you assign an offset (type x -99) on a symbol, some non-existing patterns are graphically rendered. For instance, it seems that linear segments are prolonged further than their real end.

hi, could you come up with a limited test case (data+mapfile) that exhibits this, as I can't reproduce this with what I have. Also, does this happen when using the GD renderer ?

Ok done (see attached files) It's specific to AGG + offset (no problem encountered with AGG/no_offset or GD/with_offset)

[diletant]

original linear segments (green), with offset (red), one of them is drawn straightly towards heading north (upper right side of the picture)

[diletant]

request : MAP=/ms4w/apps/gmap/htdocs/bug.map&SERVICE=WMS&REQUEST=GetMap?&VERSION=1.3.0&LAYERS=layer_1&STYLES=&FORMAT=image/png&BGCOLOR=0xFFFFFF&TRANSPARENT=True&SRS=EPSG:4326&BBOX=2.373046875,48.86471476180277,2.4169921875,48.893615361480194&WIDTH=256&HEIGHT=256

[tbonfort]

fixed in r7758

the bug happened when the first segment of a polyline was horizontal

thanks for the bugreport and testcase

[diletant]

Bug is still active, when you put 2 different offset (i.e "x -99", "y -99") I've updated all attached files for test case purposes (map, data, snapshot, request, etc).

Request :  http://localhost/cgi-bin/mapserv.exe?MAP=C:\\ms4w\\apps\\gmap\\htdocs\\bug.map&SERVICE=WMS&REQUEST=GetMap&VERSION=1.3.0&LAYERS=layer_1&STYLES=&FORMAT=image/png&BGCOLOR=0xFFFFFF&TRANSPARENT=True&SRS=EPSG:4326&BBOX=2.28515625,48.86471476180277,2.373046875,48.922499263758255&WIDTH=256&HEIGHT=256

Thks

[tbonfort]

the bug was due to some leftover integer rounding from the original gd-only code.

• fixed in trunk in r7958
• fixed in 5.2 branch in r7959

diletant: I'm crossing my fingers this will be the final fix for your issues. Your shapefile is such a mess it's a perfect testcase for bringing up rounding issues :)

[tbonfort]

there are still some leftover cases where the rendering is incorrect, reopening

[diletant]

may be a useful tip :

This issue is still active in 5.2.1 release. the wrong leftover cases might appear or disappear only by changing the size of the image or the bounding box.

Thanks & good luck !

[tbonfort]

I couldn't reproduce the defects in 5.4, 5.6 and 6.0, so I'm closing this.

reopen with an updated test-case if this is still ocurring.

thanks for the test-cases!