Alan Hourihane wrote:
Remember that bash 3.2 didn't have this problem, so it'd be worth investigating what changed.
A new asynchronous signal handling has been introduced in readline 6.0.Especially, there is a new variable named _rl_interrupt_immediately which indicates if the signal must be handled immediately or asynchronously.
I have tried the following change in lib/readline/signals.c:
static RETSIGTYPE
rl_signal_handler (sig)
int sig;
{
if (1/*_rl_interrupt_immediately || RL_ISSTATE(RL_STATE_CALLBACK)*/)
{
The behaviour seems to be OK now, but I'm absolutely not sure it is correct.
Some ideas for further investigation here:
http://osdir.com/ml/bug-readline-gnu/2009-11/msg00001.html
http://comments.gmane.org/gmane.comp.gnu.readline/43
http://www.mail-archive.com/debian-bugs-dist@lists.debian.org/msg708760.html
--
Vincent Rivière