intersect of lines and polygons resulting in duplicate linework

[DeborahVeen]

I have a line layer and a polygon layer both in the same geodatabase (file gdb in v9.3.1).  When I run an Intersect or Identity I end up with some duplicate linework (2 lines exactly on top of each other).  It appears that the duplicates are only created where the lines are perfectly coincident with the polygon boundaries.  One of the newly created duplicate lines has the attributes for the polygon on one side of the boundary and the second has the attributes of the polygon on the other side of the boundary. 

Searching the forums I found some older discussion on a similar issue when intersecting points with polygons. It appeared to be a bug that caused adjacent polygons to act like overlapping polygons when the point fell directly on the boundary between the polygons even though the polygons had clean topology and were not actually overlapping. I believe in theroy when a point or line falls directly on the polygon boundary it should be assigned the values of the polygon on the left.  But I found no other references for a problem with lines and polygons.

Have other people experienced this?  Is this a known issue? 

I'd appreciate any information!

[ChrisSnyder]

I have dealt with this for quite a long while. I don't think it is necessarily a bug, but I agree it's annoying. At the most basic level though, if you have a feature that is exactly on the border, it should probably be identified as being a member of both, right? For example, if I am right on the US/Canada border, I tend to think I am in both contries at the same time. I find this a common problem with my stream layer and my forest inventory unit layer since the stream layer was actually used to edit/split the forest unit layer (since streams seem like a pretty obvious boundary).

The easiest way to MINIMIZE the problem is to simply use the smallest possible XY tolerance in the overlay (i.e. don't use a 10ft tolerance!!!).

The other (harder but better) option is to find the duplicate features and randomly choose (or maybe using some sort of logic-based algorithm) and then delete all but one of them.

[NiklasNorrthon]

Another even better way could be to post something in ideas.arcgis.com about a optional parameter to those methods that would manage the behavior in such cases. (Only left, Only right, Random one, both, or none)

Or whatever...

Not an important issue for me, so I leave it up others to do the actual post.

[DeborahVeen]

It would be nice if we had an option on the behavior as mentioned - I will try to post that on ArcGIS Ideas. 
It does make sense that the line is identified with both polygons but because I was unaware of this happening I had been doublecounting the line miles in my project which threw off my analysis, would have been nice if it was stated in the tool help. 
I was hoping there was an easy way to find the duplicates in the attribute table (FID of -1 or some such) without having to use topology but that doesn't appear to be possible.  Has anyone found another way?
Thanks for your responses!

[ChrisSnyder]

This Python script is how I sort it out...