> Horst H. von Brand writes:
> > "Albert D. Cahalan" <acahalan@cs.uml.edu> said:
>
> >> Long term, I think the whole core file idea needs to be tossed in
> >> favor of just starting a debugger.
> >
> > Obviously also paging the party responsible for the crashed program. Using
> > some random vt for the task when a daemon crashes, too, I suppose?
>
> If crash handler registered
> stop the task and notify a registered user-space handler,
> else
> dump core.
>
> Crash handlers could be per-UID. They could listen on a device or
> expect real-time (why call it that?) signals with data attached.
> They could be full graphical GNOME apps that look up the maintainer
> in an RPM database and offer to mail a form letter. The process
> could be kept alive, which is more useful than a core. Someone with
> a clue would get a chance to start up ddd on a live process, with
> all the network connections intact.
>
> Core files just aren't likely to generate useful bug reports.
> Maybe 1% of the time someone submits a proper report.
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.rutgers.edu
> Please read the FAQ at http://www.tux.org/lkml/
I don't know about bug reports, but I have found core dumps very helpful to
locate the problem area when writing code, especially if running it in the
debugger is inconvenient.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/