It is complete outline of a trace, but it’s not showing up as a closed object. The nodes do form a simple closed path, but apparently KiCAD didn’t create them in any particular order in the file. LB isn’t seeing it closed because it can only link one vector to another in the order and direction they appear in the file.
I tried “close path” and of course it’s ignoring the obvious next vector it already shares an endpoint with, and instead tries to connect to a random vector elsewhere, over and over.
As noted in the now deleted reply, the import was unusually tiny. It was a simple matter to select all and scale it up to a workable size, but you are correct about all the pieces.
It would be mind numbing to individually grab each node and snap it to the next one, expecially as what should be a single straight line is composed of multiple un-joined segments.
I feel it would be easier to scale the original up, then use it to trace a fresh-start replication. I started to type in the suggested sequence of steps and discovered that it’s easier just to do it. The replicated file is attached and is contiguous.
The only change you might consider (or not) is to use the S in node tools to smooth the direction changes, which would convert the short straight segments to a simple curve, possibly allowing you to D delete extra curve nodes.
I cranked up the scale to make it easier to find the nodes and did not reduce it in this attached file. Be sure to check the layer settings, as I left them at your layer defaults, other than to remove the two passes.
When I imported the file, it generated a warning that this file (lbrn2) was created with v2 of the software, yet my version goes only to 1.7.07 which makes me wonder if you have an inside track? Oh, wait, there’s a pre-release version out there. In the immortal words of Emily Litella, “Never mind.” duplicated traced pcbtest2mil.lbrn2 (63.7 KB)
i fixed the scale and found sometimes the export could close, sometimes not. the order of the vectors is critical to LB but the CAD software doesn’t deliberately control the ordering in the output
i did find that when it didn’t close, i could do a small Offset on the whole thing and ended up with a closed inner and outer loop.
from that point, if you want to be exact, or at least closer to exact, you can delete one of the loops and offset the same amount back to the original dimensions. the corners may be slightly different on a very minute scale
