[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [MiNT] qed
Am 13.08.2009, 22:22 Uhr, schrieb Jo Even Skarstein <joska@online.no>:
I guess I've missed something obvious, but how can the keyboard affect
AES applications like this? AES applications get their keypresses
through the AES (the event system) and not by reading from the keyboard
device. It's the AES that does this, so if Ctrl+C does anything at all
it should kill the AES. The kernel doesn't know which application the
AES is forwarding keypresses to, so it wouldn't know which application
to kill anyway.
You're close to be right.
On ^C killgroup(SIGINT) is called by MiNT, which sends SIGINT to all
procs with the pgrp of curproc. This is ignored by XaAES, but not
necessarily by its clients. In the official XaAES all clients have the
same pgrp as XaAES, so it is possible that an XaAES-client receives
the SIGINT directly by MiNT (use the source ...).
This is all theory, I aint gonna make any tests on this, and I
dont know how N.AES handles this.
Additionally in my XaAES-version every client has its own pgrp, but for
another reason.
And a SIGINT should not cause qed to freeze.
All I know is that whithout my extra-\r the tty may go into COOKED-mode
on my system.
I suspect that Lonnys problem has another reason - buffer-overflow while
copying to clipboard somewhere. The clipboard-content can be intact, but
the data may have overwritten something that is reached at a later stage.
Theory again, but if its true, it will be hard to reproduce.
-Helmut
PS: D. Knuth once stated something like no program that you wrote only
once is of any good ... in other words: Do It Again! I think thats what
qed actually means.