On 12/01/2017 12:19, Salz, Rich wrote:It was a mix of what was done, and then a conscious decision to do things that way. As for the PR, well, maybe... We'd need to know details of which machine "test/sanitytest.c" fails on, and how popular it is to see if it's worthwhile. That would be inefficient churning given the number of changes to replace conforming null pointer initialization with memset/calloc that have gone in since this decision was made. The decision sticks in the throat a bit for us standard nerds and old-timers who remember machines where the null pointer was not all-bits-zero, but it's decades since I heard of such a machine at large in the real world. -- J. J. Farrell Not speaking for Oracle |
-- openssl-users mailing list To unsubscribe: https://mta.openssl.org/mailman/listinfo/openssl-users
