ggs@ulysses.UUCP (08/23/83)
Sounds like a fine idea, it would have eliminated the last few 'goto's that I used. And how about continue label for nested loops. I would probably use that one just for clarity in a single loop. Sometimes the object of the 'continue' gets lost in the nesting. (I know, call out the purity squad and take him away). -- Griff Smith Bell Labs, Murray Hill Phone: (201) 582-7736 Internet: ggs@ulysses.uucp UUCP: ulysses!ggs
rbk@sequel.UUCP (08/24/83)
When I was at the U. of Wisconsin and took the compiler course, we created a compiler for a language called "NOGO-III" (ie, 3rd version of a goto-less language). It supported the labelled "break" and "continue", was quite reasonable, and eliminated almost all need for gotos (in fact, since there was no goto, you were stuck if you thought you needed one). I have always been disappointed that C didn't support such a feature. Maybe it should be added to various "standards" or future versions of the language... -- Bob Beck Sequel Computer Systems ...ogcvax!sequel!rbk (503)627-9809
riks@shark.UUCP (Rik Smoody) (08/25/83)
I just don't see that a lot of gibberish with setbreak(someflag), testbreak(someflag), multibreak(someflag), and several occurrences of breakout(soemflag) is any more clear for human readers, any more efficient for computers, or any easier for a compiler to handle than a simple GOTO. The suggestion of a labelled break, on the other hand, seems to simplify and eliminate both the harm of a goto and the need to use them.
ptw@vaxine.UUCP (P. T. Withington) (08/29/83)
On the other hand, ECL was rather mind-blowing in power. I always liked to think of it as "the civilized man's Lisp". Any other fans out there? What ever became of ECL? Where have all the flowers gone? Etc., etc.
jsq@ut-sally.UUCP (08/30/83)
> (CAUTION) > > Labeled breaks sound reasonable, but leave it at that please. >I had the excruciatingly unpleasant experience of working in a language >called ECL (or EL1 -- your choice) at Harvard. Among other unpleasantries >the language supported labeled blocks that you could break out from anywhere >within the block (even from subroutines called within the block) with >some command like "=> <label>". The language was written in such a way >that you were occasionally forced to use this construction -- and it >was a roaring pain to debug. > >Craig Partridge >cfib!craig The language was called EL/1, the support package was called ECL. It's all in what you're used to: I found the language quite easy to use and debug. Certainly easier than tracing down longjmps and gotos. -- John Quarterman, CS Dept., University of Texas, Austin, Texas {ihnp4,ut-ngp}!ut-sally!jsq, jsq@utexas-780.ARPA (soon to be jsq@ut-sally.ARPA)
craig@cfib.UUCP (08/30/83)
#R:ulysses:-57000:cfib:5900003:000:568 cfib!craig Aug 26 15:05:00 1983 (CAUTION) Labeled breaks sound reasonable, but leave it at that please. I had the excruciatingly unpleasant experience of working in a language called ECL (or EL1 -- your choice) at Harvard. Among other unpleasantries the language supported labeled blocks that you could break out from anywhere within the block (even from subroutines called within the block) with some command like "=> <label>". The language was written in such a way that you were occasionally forced to use this construction -- and it was a roaring pain to debug. Craig Partridge cfib!craig
preece@uicsl.UUCP (08/31/83)
#R:ulysses:-57000:uicsl:6400006:000:701 uicsl!preece Aug 30 11:13:00 1983 The SAIL language allows the BEGIN and END 'statements' to be followed by string constants. Those constants are then matched by the compiler, making missed delimiters very easy to identify and fix, and can be used in break and continue statements. Another thing that would be very useful in case statements would be a 'recase' verb, which would restart the enclosing switch (i guess in C it would be 'reswitch'). It is often useful to be able to recognize some cases as only minor variants of others, and process them by using a prologue for one variant followed by the existing case treatment shared between them. A reswitch verb would make this trivial. scott preece pur-ee!uiucdcs!uicsl!preece
alan@allegra.UUCP (08/31/83)
Another thing that would be very useful in case statements
would be a 'recase' verb, which would restart the enclosing
switch (i guess in C it would be 'reswitch'). It is often useful
to be able to recognize some cases as only minor variants of
others, and process them by using a prologue for one variant
followed by the existing case treatment shared between them.
A reswitch verb would make this trivial.
In certain cases, C allows you to do exactly what you're asking for.
switch (a) {
case ONE_THING:
/* code for ONE_THING */
break;
case A_VARIATION_ON_A_SECOND_THING:
/* prologue, then fall through... */
case A_SECOND_THING:
/* common code for A_SECOND_THING and variation */
break;
}
Of course, there can't be two variations, each with its own prologue,
of one case, so this is of limited use.
Alan Driscoll
Bell Labs, Murray Hillguy@rlgvax.UUCP (Guy Harris) (09/01/83)
"goto"s can be used for treating several cases as subcases of one common
case (i.e., do the case-specific thing and then the common actions), generally
without too tangled code. Furthermore, if it *does* tangle things, or you don't
want to use "goto"s, many versions of the C "final" optimizer will do the right
thing if you write something like
case (foo) {
case 1:
/* 1-specific things */
/* common code */
break;
case 2:
/* 2-specific things */
/* common code */
break;
.
.
.
and join the common code and stick a jump/branch instruction into the cases.
(Versions known to: PDP-11, VAX-11, Zilog "Zeus", CCI - and probably most
other - 68000).
Guy Harris
{seismo,mcnc,we13,brl-bmd,allegra}!rlgvax!guycraig@cfib.UUCP (09/02/83)
#R:ulysses:-57000:cfib:5900004:000:400 cfib!craig Sep 1 11:31:00 1983 In response to John Quarterman... Indeed, ECL is the support package and EL/1 is the language -- people at Harvard rarely bother to distinguish anymore. I'm more surprised to hear that you actually liked programming in the language. It puts you in a rare class -- most people I know despise it, and indeed, try to avoid the two courses at Harvard that still use it. Craig Partridge cfib!craig