All compiles except where marked. All should compile

int main()
{
    char        c;
    char        *p  =   &c;
    unsigned    x   =   0;
    char        b[10];

    p[x] = 'o';
    x[p] = 'k';

    0[p] = 'r';

    b[x] = 'o';
    x[b] = 'k';     /* Compiler wrongly gives error */

    0[b] = 'r';     /* Compiler wrongly gives error */

    return 0;
}

Matthew wrote:
> All compiles except where marked. All should compile

i think it may better stay so. It's definately not very annoying since anyone sane wouldn't write that. You could just as well write (x+b) :>

-eye

> Matthew wrote:
> > All compiles except where marked. All should compile
>
> i think it may better stay so. It's definately not very annoying since anyone sane wouldn't write that. You could just as well write (x+b) :>

Nonsense. It's non-standard behaviour, and in such a simple thing that any other compiler handles correctly. I really don't think you'll convince Walter to *not* do this, given his desire to be fully-conformant.

And as for your assertion that no-one would use such a construct, I'll just smile, and you can wait until you've read "Imperfect C++". Now you've got three reasons to buy it (the other ones being the Properties chapter, and my mentioning your request for properties in the acknowledgements).

:)

Matthew wrote:
>>Matthew wrote:
>>
>>>All compiles except where marked. All should compile
>>
>>i think it may better stay so. It's definately not very annoying since
>>anyone sane wouldn't write that. You could just as well write (x+b) :>
> 
> Nonsense. It's non-standard behaviour, and in such a simple thing that any
> other compiler handles correctly. I really don't think you'll convince
> Walter to *not* do this, given his desire to be fully-conformant.

It's probably so to guard for bugs.

> And as for your assertion that no-one would use such a construct, I'll just
> smile, and you can wait until you've read "Imperfect C++". Now you've got
> three reasons to buy it (the other ones being the Properties chapter, and my
> mentioning your request for properties in the acknowledgements).

I really don't need to count the reasons, i know i want to get it as soon as it's out. :> Unfortunately, /me would not be able to raise your sales much only by bying it myself.

-eye

"Ilya Minkov" <minkov@cs.tum.edu> wrote in message news:c169kc$1fdf$2@digitaldaemon.com...
> Matthew wrote:
> >>Matthew wrote:
> >>
> >>>All compiles except where marked. All should compile
> >>
> >>i think it may better stay so. It's definately not very annoying since anyone sane wouldn't write that. You could just as well write (x+b) :>
> >
> > Nonsense. It's non-standard behaviour, and in such a simple thing that
any
> > other compiler handles correctly. I really don't think you'll convince Walter to *not* do this, given his desire to be fully-conformant.
>
> It's probably so to guard for bugs.
>
> > And as for your assertion that no-one would use such a construct, I'll
just
> > smile, and you can wait until you've read "Imperfect C++". Now you've
got
> > three reasons to buy it (the other ones being the Properties chapter,
and my
> > mentioning your request for properties in the acknowledgements).
>
> I really don't need to count the reasons, i know i want to get it as soon as it's out. :> Unfortunately, /me would not be able to raise your sales much only by bying it myself.

Don't be so sure. You might account for up to 50% of sales! (I know my mum's
going to buy one <g>)

Matthew wrote:
>>I really don't need to count the reasons, i know i want to get it as
>>soon as it's out. :> Unfortunately, /me would not be able to raise your
>>sales much only by bying it myself.
> 
> Don't be so sure. You might account for up to 50% of sales! (I know my mum's
> going to buy one <g>)

I've been spreading Word, so i'm sure i account for at most 34% of sales.

-eye

"Matthew" wrote...
> All compiles except where marked. All should compile
<snip>

Have Walter fixed this already??? :-o Just tried it with 8.40 and it compiles fine...

A very impress person,
KTC

-- 
Experience is a good school but the fees are high.
    - Heinrich Heine

He usually posts a response on the NG.

Did you try all the cases?

I just tried with 8.40.1, and it still fails.

"KTC" <me@here.com> wrote in message news:c18pf7$30pl$1@digitaldaemon.com...
>
> "Matthew" wrote...
> > All compiles except where marked. All should compile
> <snip>
>
> Have Walter fixed this already??? :-o Just tried it with 8.40 and it compiles fine...
>
> A very impress person,
> KTC
>
> -- 
> Experience is a good school but the fees are high.
>     - Heinrich Heine
>
>

"Matthew" wrote...
> He usually posts a response on the NG.
>
> Did you try all the cases?
>
> I just tried with 8.40.1, and it still fails.
>

I've just copied and paste exactly what you posted in the orginial post and it compiles fine for me...

scppn.exe gives version 8.40.2n

KTC

-- 
Experience is a good school but the fees are high.
    - Heinrich Heine

>
> "Matthew" wrote...
> > He usually posts a response on the NG.
> >
> > Did you try all the cases?
> >
> > I just tried with 8.40.1, and it still fails.
> >
>
> I've just copied and paste exactly what you posted in the orginial post and it compiles fine for me...
>
> scppn.exe gives version 8.40.2n

That's what I have.

I must conclude that I pasted an incomplete version. Including here again ...

Walter, I presume this'll be easy-ish to fix. Any ETA? :)