weird behaviour of readline on OSX 10.6
You seems to use Ruby 1.8.x.
So can you try following?
--- ext/readline/readline.c (revision 26664)
+++ ext/readline/readline.c (working copy)
@@ -833,6 +833,12 @@
rl_event_hook = readline_event;
- rl_catch_signals = 0; +#endif +#ifdef HAVE_RL_CATCH_SIGWINCH
- rl_catch_sigwinch = 0; +#endif #ifdef HAVE_RL_CLEAR_SIGNALS rl_clear_signals(); #endif Index: ext/readline/extconf.rb =================================================================== --- ext/readline/extconf.rb (revision 26664) +++ ext/readline/extconf.rb (working copy) @@ -59,6 +59,9 @@ have_readline_var("rl_attempted_completion_over") have_readline_var("rl_library_version") have_readline_var("rl_event_hook") +# workaround for native windows. +/mswin|bccwin|mingw/ !~ RUBY_PLATFORM && have_readline_var("rl_catch_sigwinch") +/mswin|bccwin|mingw/ !~ RUBY_PLATFORM && have_readline_var("rl_catch_signals") have_readline_func("rl_cleanup_after_signal") have_readline_func("rl_clear_signals") have_readline_func("rl_vi_editing_mode")
(2010/02/19 4:59), Andrew Eberbach wrote:
I've been noticing that irb (and as a result script/console in rails)
don't behave correctly with Control C. Nothing would happen when I hit
control C until I hit enter or a few more keys and then it would clear
the line and show a C and reset to the prompt. This doesn't happen on
Linux and it didn't happen on 10.5
If there's a long running process in irb hitting control c works as
expected but not if it's just sitting at a prompt.
I played around with the ext/readline.c and found that if I put a
rl_catch_signals = 0;
Then everything works as normal. Should this be added as a patch?
Reading the readline docs it seems to me that what's happening is the
internal readline handler gets stuck sending SIGINT back to the ruby
process even though there's a trap("SIGINT") defined. Any ideas?
NARUSE, Yui firstname.lastname@example.org