reporting R crashes as bugs

Recently I am crashing R relatively often. The message is invariably

*** caught segfault ***
address xxx, cause 'memory not mapped'

Reproducing would be hard, and in any case I could not share the large proprietary data sets. My que开发者_StackOverflow中文版stion is: can I report the crash so that it is investigated? Would generating the coredump help? It seems that the FAQ section on reporting bugs is not meant for this contingency.

---

I wouldn't report them to R-core / R-devel unless you're 100% certain it's a bug in R and not a package. It's more likely the bug is in a package you're using. If you can't create a reproducible example of the bug, the package maintainer isn't likely to try to hunt it down.

---

This error message can occur with

1. Hardware overheating
2. Using outdated versions of R or of packages
3. A bug in a package
4. A bug in R

Until you can reproduce it, with fake data if you don't want to expose your own, that I doubt anyone will pay attention to a crash report.

---

The 'memory not mapped' text leads to a strong suspicion of this being a systems issue: either one of OS, C library or BLAS could be at stake.

It is frustrating with large compute jobs -- but there is no other way than to distill it to a minimal reproducible example, which you could do with mocked-up data too.

As a start, try to see if the problem persists on a different OS, eg try re-running it on a mac or a different linux flavour or windoze.