On Tue, Mar 19, 2019 at 5:10 PM Martin Buchholz <marti...@google.com> wrote:
> I'm the one who introduced vfork for process spawning, and I support > Thomas' plan to switch to posix_spawn. > (although I have not seen a crash attributed to using vfork on Linux) > Thanks Martin. I think for a long time vfork was the best alternative. I remember seeing one or two cases in the last five years which could have been caused by vfork. In both cases, in a fork-heavy environment with many short-lived processes, a broken script killed the wrong pid - a just spawned child - and pid's father disappeared at the same time without a trace. Cheers, Thomas
