Me: *send crashlog*
Support: Could you send your system spec and Houdini diagnostic info?
Me: Ok *send said info*
Support: Could you send the .hip file that crashed?
Me: Ok *send said file*
Support: Sorry we couldn't reproduce it, and we can't fix unreproducible issues.
Me: Ok *sad face*
Conversation ended because there was nothing left to do.
Yeah, of course you can't reproduce it. I couldn't reproduce it either! It was a random crash! As anyone who wrote C/C++ knows, crashes typically happen because you use freed memory, and it depends on the whole series of instructions you did, the memory layout etc. Most crashes are not easily reproducible (otherwise they would have been fixed, of course).
Now when my Houdini crashes, I don't know whether I should bother to report. (or should I even keep using Houdini...) I know the support will just repeat the above conversation. It's not his/her fault tho. What could they do? There must be a policy preventing them from adding "random crashes" to their issue tracking system, and it's justified: otherwise their dev's queue will be flooded with unfixable bugs.
I don't know how to improve this situation. But lately I encountered an issue with Rider (an IDE). This is their response:
Maybe it's what Houdini lacks? A way to collect logs for debugging.