Path: csiph.com!eternal-september.org!feeder.eternal-september.org!reader01.eternal-september.org!.POSTED!not-for-mail
From: Keith Thompson <Keith.S.Thompson+u@gmail.com>
Newsgroups: comp.lang.c++
Subject: Re: two's complement idea
Date: Mon, 27 Jan 2020 19:15:58 -0800
Organization: None to speak of
Lines: 73
Message-ID: <874kwg46fl.fsf@nosuchdomain.example.com>
References: <h41XF.367918$Arx.178294@fx08.am4> <r0jumc$p8e$1@dont-email.me> <mQfXF.125675$Bf2.15315@fx39.am4> <r0k5um$375$1@dont-email.me> <%JhXF.102988$%G5.101221@fx03.am4> <r0kkn9$t6q$1@dont-email.me> <49pr2fphchn3dk2hiqv2bk8ml5oijs0d6e@4ax.com> <DZmXF.242725$_R5.99089@fx19.am4> <p3ks2f1vekfr0dckvguuvkrdr6lrpedrsl@4ax.com> <VhGXF.46342$BA5.20568@fx21.am4> <e1hu2f5p960nhg1spsvndon031nop7ldhr@4ax.com> <r0njph$86d$1@dont-email.me> <oLIXF.113152$%G5.81633@fx03.am4> <go3v2f9mpoe2shqdvvakdrp4rb6n9bl6e9@4ax.com>
Mime-Version: 1.0
Content-Type: text/plain
Injection-Info: reader02.eternal-september.org; posting-host="3b9d7e67aee25720b85e28e193eb2d8c"; logging-data="1786"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX19hjk2IoQ7JYdUDzr0oO7nY"
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/26.3 (gnu/linux)
Cancel-Lock: sha1:qOUfiNQB1DNUmWdNInw96+kR18c= sha1:kvcG/JlH2Cx7eFgyog8bCvG+g4Y=
Xref: csiph.com comp.lang.c++:71634

Robert Wessel <robertwessel2@yahoo.com> writes:
> On Mon, 27 Jan 2020 21:38:26 +0000, Bart <bc@freeuk.com> wrote:
[...]
>>That makes it harder (would have taken longer than the 20 minutes I'd 
>>set aside). And it raises one issue: if this label is also a legal goto 
>>label, then:
>>
>>     outerloop:
>>     for(;;) { while(1) {
>>         goto outerloop;
>>         break outerloop;
>>     }}
>>//  endofouterloop:
>>
>>the statements will jump to completely different points. The break 
>>actually goes to an implicit statement after the loop.
>
>
> I think the slight oddity of target location when the label is used
> for both a break/continue and a goto is pretty minor.  I suppose we
> could come up with some alternative way to specify a "statement name",
> but I don't see a big advantage.  If we can have a dozen meanings of
> static, why not this.
>
>>At this point, it seems to me the following would actually be clearer:
>>
>>     for(;;) { while(1) {
>>         goto endofouterloop;
>>     }}
>>     endofouterloop:
>>
>>as it pinpoints exactly where the code will end up next. With loads of 
>>nested }s floating around, many nothing to do with the loops, the end of 
>>the desired loop is not always easy to find.
>
> I think that's the wrong principle.  If you want to target a specific
> location, use a goto.  The idea is to break/continue a loop/switch, so
> it should be the *loop/switch* that should be identified.

Personally, I like the way Ada does this.  You can apply a label to a
loop, and it's the name of the loop ("--" introduces a comment):

    OUTER: while condition loop
        INNER: while condition loop
            -- do stuff
            if Done_With_Outer then
                exit OUTER; -- "exit" is Ada for "break"
            elsif Done_With_Inner then
                exit INNER;
            end if;
            -- do more stuff
        end loop INNER;
    end loop OUTER;

"exit OUTER" is best thought of as an operation that applies to the loop
named "OUTER" rather than as a branch to a particular location.

Loops that are the targets of goto statements have a different syntax,
one that was designed to stand out:

    <<LABEL>>
    ...
    goto LABEL;

On the other hand, Perl does something similar, but uses the same syntax
for loop names and goto labels.  This does mean that seeing a named loop
raises the possibility that somewhere there's a goto that targets it,
but that's not much of a problem in practice.

-- 
Keith Thompson (The_Other_Keith) Keith.S.Thompson+u@gmail.com
Working, but not speaking, for Philips Healthcare
void Void(void) { Void(); } /* The recursive call of the void */