I've never liked the approach Linux kernel and userland take to memory exhaustion. Many people confidently asserts that it never happens. The somewhat better-informed suggest that it's unreasonable to write programs that recover from memory exhaustion because unwinding requires allocation --- a curious belief, because there are many existence proofs of the contrary. Then we get a feedback loop where everyone uses overcommit because everyone believes that programs can't recover from OOM, and people avoid writing OOM recovery code because they believe that everyone is using overcommit and allocation failure is unavoidable. And then they write kernel code and bring this attitude there.
Memory is just a resource. If you can recover from disk space exhaustion, you can recover from memory exhaustion. I think the current standard of memory discipline in the free software world is inadequate and disappointing.
Memory is just a resource. If you can recover from disk space exhaustion, you can recover from memory exhaustion. I think the current standard of memory discipline in the free software world is inadequate and disappointing.