Import Features

[okaragoz]

Is it possible to include import features for geo-package or shapefile vector files? I tried to replace "input --> layer limits" with my existing vector files (gpkg and shp) with 6000 polylines, and QGIS crashes when I clicked reproduce map.

The specs are Ryzen 9 5900X, 32 Gb Ram, Nvme SSD, Python 3.9, and QGIS 3.22.

[luca-penasa]

Hi @okaragoz

It should work without too many issues also with your own files.

This is probably due to the dense dataset you are processing. We never tested with so many lines. Anyway, the polygon generation should work within a reasonable time. What is likely to stuck the computation is the generation of clean lines, which implementation is currently super-slow.

If you are using the new interface try disabling that option.

If you are willing to share your dataset, or a similar one I'd be happy to see if we can do anything about it.

[okaragoz]

Hi @luca-penasa Thanks for the reply. Unfortunately, it is not possible to share the dataset, however, I have experienced the same problem when I do a test with other vector datasets. For example, I tested the Tanaka 2014 (SIM3292) polygon shapefile and I converted it to polylines (with closed line boundaries). I export it to gpkg and replaced it with the "input --> layer limits" dataset. QGIS crashes whenever I clicked reproduce map.

PSI: The new interface is disabled and I don't think there is a problem with the file size.

[luca-penasa]

Following up on this. I tried to use the tanaka dataset with mappy and here are some points for easier import existing mapping projects:

• transforming polygons to meaningful lines is not trivial and to my current understanding, the only qgis tool suitable for this task is the v.to.lines tool from GRASS.
• In the latest release of the plugin (available on github only for now) I added an experimental Import polygonal map algo that tries to use grass to import the project. But it would require testing and feedback (and requires grass7 algorithms).
• the only problem is that the output lines are broken at each intersection, which might be suboptimal. But there is no obvious way to determine the author's intention when drawing the line (an interesting way would be to join back all the lines which appear to be ~straight at the intersection).
• the tanaka map is regenerated from lines and points in around 4-5 secs which might be too much but makes the tool still usable.
• the algorithm taking care of removing dangling ends is instead too slow and would require a complete rewrite. For now, the best solution would be probably to explore what grass has to offer for that task (v.clean should provide everything needed).