Re: Gleef's comments
- From: Gleef <gleef capital net>
- To: Chris Jantzen <chris maybe net>
- cc: gnome-gui-list gnome org
- Subject: Re: Gleef's comments
- Date: Sun, 9 Aug 1998 10:40:38 -0400 (EDT)
On Sat, 8 Aug 1998, Chris Jantzen wrote:
> Gleef writes:
> > I do remember that I once said that a program should treat SIGINT and
> > SIGTERM as if the user hit Exit, and treat SIGKILL quickly and
> > non-destructively, with no user input. Not the same thing as he's
> > quoting, but I don't remember anyone else talking about signals at all.
>
> A user application can't trap SIGKILL. Your program dies. Period. :-)
> If this were not so, rogue/trojan applications would be much easier to
> write under UNIX.
Sounds reasonable. In that case a program should (C2 or C3) be designed
so that a SIGKILL does not cause a catastrophic loss of data. For an
example, we can use Emacs, which uses temporary save files named
"#filename#" and removes them when done. This allows an Emacs process to
disappear, leaving the original, a backup "filename~" file (if present),
and the working "#filename#" file.
We should either require or suggest that GNOME programs follow behavior
along similar lines. I suspect there will be too many exceptions for
requiring it to make sense, so it's probably a C3 (or whatever the level
is called today).
-Gleef
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]