Bugs item #764685, was opened at 2003-07-02 15:52 Message generated for change (Comment added) made by simonmar You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=108032&aid=764685&group_id=8032
Category: None Group: None >Status: Closed >Resolution: Fixed Priority: 5 Submitted By: Hampus Ram (hampusr) >Assigned to: Simon Marlow (simonmar) Summary: GHCi breaks the terminal Initial Comment: After running ghci the terminal does not work exactly as it should: Running "su" does no longer work for some unknown reason (su stops after reading the first character of the password). This happens with ghc 5.04 as well as 6.0 and current CVS-version, both in xterm and console. This only seem to be a problem with my Linux/Debian installation, I cannot reproduce the problem on Solaris. Also some minor spelling :) According to the CVS users-guide the reverse of -frules-off is the very same -frules-off. ---------------------------------------------------------------------- >Comment By: Simon Marlow (simonmar) Date: 2003-09-24 10:36 Message: Logged In: YES user_id=48280 Fixed, thanks. It turns out that it is nothing to do with readline, but indeed GHC's IO library should really be restoring the terminal settings if it changes them. Note that #811427 is another report of the same bug. ---------------------------------------------------------------------- Comment By: Hampus Ram (hampusr) Date: 2003-07-28 10:20 Message: Logged In: YES user_id=159064 I performs a ioctl at the end, strace says: ioctl(1, SNDCTL_TMR_TIMEBASE, {B38400 opost isig icanon echo ...}) = 0 when running with input from /dev/null. This is all ok. However when running in the terminal it says: ioctl(1, SNDCTL_TMR_TIMEBASE, {B38400 opost isig -icanon echo ...}) = 0 Not so good. Dpkg reports that I'm using readline 4.3-5. I'm running bash (2.05b-7) as my primary shell but I just tried tcsh and there it all works well! ---------------------------------------------------------------------- Comment By: Simon Marlow (simonmar) Date: 2003-07-28 09:20 Message: Logged In: YES user_id=48280 Readline is supposed to reset the terminal. Could you try strace ghci </dev/null and see if it attempts to reset the terminal at the end? My trace has an ioctl(1, ...) near the end. What version of readline? Is it different from your Debian- running friends? What shell are you using? (some shells save/restore the terminal settings between running commands). ---------------------------------------------------------------------- Comment By: Hampus Ram (hampusr) Date: 2003-07-27 21:27 Message: Logged In: YES user_id=159064 The (only) difference is that instead of icanon (before) ghci leaves the terminal with -icanon (su no longer works using line buffering, I think). I've had this problem before with an older installation of Debian too but no one else of my Debian-running friends seem to have this problem. Well, it's a minor problem but annoying. ---------------------------------------------------------------------- Comment By: Simon Marlow (simonmar) Date: 2003-07-22 11:27 Message: Logged In: YES user_id=48280 I can't reproduce this on our boxes here. It could be a readline problem. Could you try 'stty -a' before and after running GHCi, and post the results? ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=108032&aid=764685&group_id=8032 _______________________________________________ Glasgow-haskell-bugs mailing list [EMAIL PROTECTED] http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs