I have been working through figuring out why we are getting a consistent hang on t5562.08 for 2.28.0 on the NonStop TNS/X platform (x86 big endian). This did not happen in the -rc series changes and appears transient. My investigation into this did not turn up anything useful other than the circumstances. When run from a shell, the test consistently passes regardless of options used. Then only time this hangs is when we run through the git test suite through Jenkins ? so there is no valid input tty. Even redirecting stdin to /dev/null does not cause an issue. I am suspecting something in the test suite rather than git itself. We have had similar issues with t5526 in the past ? 2.20 timeframe as I recall. Interrupting git-http-backend unwound the test and allowed us to continue with the pipeline but I have no evidence of anything wrong there. This will not stop our release of the NonStop builds of 2.28.0 since the test passes in isolation. Curiously, this does not have an issue on the TNS/E platform (itanium big endian) where we have seen more issues historically. Kind Regards, Randall -- Brief whoami: NonStop developer since approximately 211288444200000000 UNIX developer since approximately 421664400 -- In my real life, I talk too much.
