On Mon, 20 Jan 2025 17:13:19 GMT, Roger Riggs <rri...@openjdk.org> wrote:
> In ProcessImpl_md.c, we have adopted the Posix APIs and semantics for process > handling. I suggest removing the "unofficial" link, it does not include > useful information at this point and is fragile. > > The current behavior always setting SIGCHLD to SIG_DFL is/has been reliable > and does not depend on the behavior of the parent process. (Good) @tstuefe > May also be interested in reviewing this change. https://pubs.opengroup.org/onlinepubs/9799919799/ seems to indicate that the "default action is to ignore the signal", which is unclear to me. I think it means that the signal will not cause invocation of a user handler or abort the process, but does this also mean that wait and waited won't work? In any case, I agree with @RogerRiggs that keeping this behavior is fine. > > The previous sentence mentions Solaris and that should be removed at some > point but out of scope for this PR. ------------- PR Comment: https://git.openjdk.org/jdk/pull/21633#issuecomment-2603125595
