Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!uunet!husc6!necntc!ima!haddock!karl
From: karl@haddock.ISC.COM (Karl Heuer)
Newsgroups: comp.lang.c
Subject: setjmp, longjmp, goto
Message-ID: <990@haddock.ISC.COM>
Date: Fri, 21-Aug-87 18:34:16 EDT
Article-I.D.: haddock.990
Posted: Fri Aug 21 18:34:16 1987
Date-Received: Sun, 23-Aug-87 06:06:41 EDT
References: <2196@xanth.UUCP>
Reply-To: karl@haddock.ima.isc.com (Karl Heuer)
Organization: Interactive Systems, Boston
Lines: 13
Summary: If it's a builtin, maybe it should look the part?

In article <2196@xanth.UUCP> kent@xanth.UUCP (Kent Paul Dolan) writes:
>[Symbolics machines require that setjmp be a builtin]

There's an interesting connection between this and my earlier ramblings about
label variables and pointers.  If labels were real objects (with declarations,
etc.), perhaps "long label" could be a variant that includes the stack info
for non-local gotos.  Then "jmp_buf" becomes "long label", "if (setjmp(j))"
becomes "come_from = 0; j: if (come_from)", and "longjmp(j, x)" becomes
"come_from = x; goto j".

Naaaah.  (Maybe if I were designing a new language from scratch...)

Karl W. Z. Heuer (ima!haddock!karl or karl@haddock.isc.com), The Walking Lint