On Wed, 17 Mar 2021 19:13:24 GMT, Roger Riggs <rri...@openjdk.org> wrote:
>> Hmm. That's strange. >> >> One issue I don't understand with this test, should the parent not read from >> both streams simultaneously, because if the child is sending output to the >> stream the parent is not reading from the child may block on write? >> >> This depends on file io buffer size of course but if the child writes some >> larger output into e.g. stderr while the parent reads stdout, parent will >> wait on stdout and child will wait on the write to stderr finishing. So, >> either we should read both streams or redirect the stream we are not >> interested in to paren'ts stdout/stderr where hopefully the testrunner >> itself would read it. >> >> I may misunderstand the test completely; if yes, sorry for the confusion! >> >> Thomas > > The child does no (zero) writes to either stream. It is invoked only to > sleep until it is destroyed. > The purpose of the test is to verify the exception that is thrown when the > other end(child) of the pipe is closed (because the process has been forcibly > terminated). Hmm, maybe the child can create a file to indicate that bootstrap has finished. ------------- PR: https://git.openjdk.java.net/jdk/pull/3049
