At Fri, 12 Jul 2002 11:55:32 +0000 (UTC), "Ben Harris" wrote:
> This is likely to change such that abort():
>  + May be called without restriction from a signal handler, and
>  - Need not fclose() open stdio streams.
> 
> Would this change actually inconvenience anyone in the real world?  Is
> that inconvenience worth it?

Would this result in needing to explicitly flush said streams before
calling abort(), if they used buffering?

That would be unfortunate, as the error message preceeding the abort
may have been written to a buffered stream...  (I know, i know, error
messages should go to stderr and it is by default unbuffered, but not
all sw is written that way.)

If the standardized definition of abort() changes in that way, I think
we should still make the attempt to fclose() (or perform similar
operations, e.g. flushing, before closing).


cgd