Definitely a good idea.I was going to say that the assert failure routine probably flushes all buffers, but of course we could also segfault.
John On 5/25/11 11:55 AM, Eric Eide wrote:
Eric> To reproduce this bug, we need to know two things: Eric> Eric> + what version of Csmith you are running Eric> + the seed that Csmith chose that led to the crash Eric> Eric> You should be able to get both of these from output file Eric> ("text.c"). Can you provide us with those things? Note to myself (and my colleagues): we should flush the output stream immediately after printing these things, to ensure that this data makes it to disk. I will make this change.