Possible infinite loop in ArrangementBuilder

[Ylannl]

On rare occasions the ArrangementBuilder can get stuck in a seemingly infinite loop.

[bertt]

I have a case of infinite loop, difficult to reproduce. But the trace logs are:

[2024-12-21 08:09:14.289286015] TRACE   {\"name\":\"serialize\",\"count\":47}
[2024-12-21 08:09:24.289425590] TRACE   {\"name\":\"rss\",\"count\":368812032}
[2024-12-21 08:09:24.289445047] TRACE   {\"name\":\"crop\",\"count\":47}
[2024-12-21 08:09:24.289448353] TRACE   {\"name\":\"reconstruct\",\"count\":47}
[2024-12-21 08:09:24.289451118] TRACE   {\"name\":\"sort\",\"count\":47}
[2024-12-21 08:09:24.289453814] TRACE   {\"name\":\"serialize\",\"count\":47}
[2024-12-21 08:09:34.289587676] TRACE   {\"name\":\"rss\",\"count\":368812032}
[2024-12-21 08:09:34.289609507] TRACE   {\"name\":\"crop\",\"count\":47}
[2024-12-21 08:09:34.289612753] TRACE   {\"name\":\"reconstruct\",\"count\":47}
[2024-12-21 08:09:34.289615528] TRACE   {\"name\":\"sort\",\"count\":47}
[2024-12-21 08:09:34.289618113] TRACE   {\"name\":\"serialize\",\"count\":47}

[Ylannl]

Thanks for reporting. Could you share the input data and config?

[bertt]

It's very difficult to reproduce (occurs only after 3 hours processing at exactly the same place). For now I work around it by using a GeoPackage file instead of PostgreSQL as input roofprint table. Somehow this works better.

[bertt]

A more complete log of this issue in Paris, with las file LHD_FXX_0648_6845_PTS_O_LAMB93_IGN69.copc.laz. Couldn't reproduce it though :-(

france_batiments_aoi.zip
roofer_infinite_loop.log