>Hmm, right.  So if all of these are valid, I guess the question is
>which one is the most sensible action.  Crashing the app is somewhere
>between rm -rf and printing "(null)".  While I agree that it's useful
>for developers if the app crashes in this situation, it's less so for
>users.  One solution is writing perfect code, but we all know that it
>doesn't exist.


I would argue that crashing the app is more useful than printing
(null).  Clearly, the application is broken and undefined behaviour is
invoked.  Also, th eapplication turns out not to be portable.

Now, if an application invokes undefined behaviour, it is not a question
of which undefined behaviour is best, but that the *worst* case
could easily be "rm -rf /".

Apparently, not crasing has led GNOME developers to right buggy code which
crashes on Solaris.  So was it really the right thing to do in glibc?

Casper
_______________________________________________
opensolaris-code mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/opensolaris-code

Reply via email to