On Sun, Oct 18, 2009 at 09:24:58PM -0400, Taylor R Campbell wrote: > You can enter an ordinary REPL, in the debugger frame's environment, > with the `e' command.
Yes, but if the debugger could offer REPL and debug features at the same time, then "e" would become unnecessary except perhaps in case of debugger errors. Basically I'm saying that the abstractions and tools MIT Scheme provides are useful but the keystrokes to use them may be able to be improved. > Once control enters the reader, killing the line may prove tricky. > When interacting with the usual REPL, the only way one can kill a line > is by talking with the terminal before the line gets to Scheme. Talking with the terminal is fine. What I meant is that the terminal kill character should also kill the parenthesis. In order for that to happen, the parenthesis must be echoed after the prompt and the TTY driver must know the parenthesis was echoed. Currently in (%prompt-for-expression): - the output half of the terminal is put into cooked mode - the prompt is printed - the input half of the terminal is put into cooked mode - (read) is called I just wrote a debugger command that does the unread-char, and that part works fine. The reader sees the parenthesis. Doing the unread-char doesn't cause any echo, however, so unread-char isn't the solution by itself. In OS X, there's an ioctl called TIOCSTI, "simulate typed input". In a little C test program I wrote, in cooked mode the ioctl echos whatever character you want and puts it into the TTY driver's input line for erase, kill, and word-erase to work on. That's the operation I need, but the MIT Scheme microcode doesn't know about TIOCSTI yet. -- Derek _______________________________________________ MIT-Scheme-devel mailing list [email protected] http://lists.gnu.org/mailman/listinfo/mit-scheme-devel
