It's likely a 32-bit arm unaligned memory access issue. Can you check if it will also crash for a small max and ts ?
On Fri, 7 Jan 2022 at 3:39 AM Jan-Pieter Jacobs <[email protected]> wrote: > I was trying to debug my implementation of Dijkstra's algorithm and ran > into this J crash: > > (AoC Day 15 spoiler ahead -- well sort off, the solution is not entirely > correct, hence the debugging :p) > > ts is a snaking path of 0's from 0 0 to 9 9, and try1 (should) implement > Dijkstra's algorithm for finding a path with a minimal sum of the nodes > crossed. Something in my implementation is wrong (hints welcome), but the J > bug turned up when I wanted to debug this approach by adding a \ to keep > intermediate results, which causes a bus error: > > try=: [: ([ + 2 <./\ padrifshort)&.>/\@|. [: </. 0 (<0 0)} ] > ts =: 0 1 0 0 0 1 0 0 0, (7 9$9$0 1), (0 0 0 1 0 0 0 1 0) > try1=: [: ([ + 2 <./\ padrifshort)&.>/@|. [: </. 0 (<0 0)} ] > max=: 2e9 > padrifshort=: ]`(max (,,[) ])@.(>&#) > try1 ts NB. "works" no crash, but result is wrong. > ┌─┐ > │2│ > └─┘ > try ts > Bus error > NB. J crashed. > > For an explanation of the code, see ( > http://www.jsoftware.com/pipermail/programming/2022-January/059538.html). > > My J version: > JVERSION > Engine: j903/j32/android > Release-a: commercial/2021-12-17T11:10:19 > Library: 9.03.08 > Platform: Android 32 (armeabi-v7a) > Installer: unknown > InstallPath: /mnt/sdcard/Android/data/com.jsoftware.j.android/files > Contact: www.jsoftware.com > > Jan-Pieter > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
