Path: csiph.com!eternal-september.org!feeder.eternal-september.org!mx02.eternal-september.org!.POSTED!not-for-mail
From: Keith Thompson <kst-u@mib.org>
Newsgroups: comp.lang.c
Subject: Re: sizeof on an uninitialized pointer to a VLA
Date: Sat, 21 Nov 2015 12:09:40 -0800
Organization: None to speak of
Lines: 55
Message-ID: <lnd1v36rnf.fsf@kst-u.example.com>
References: <5e03af2c-3c3e-4432-bf5a-000867e5febb@googlegroups.com> <n2ljik$iqo$1@dont-email.me> <ln610wevm8.fsf@kst-u.example.com> <87mvu8rbor.fsf@bsb.me.uk> <lnsi40axv9.fsf@kst-u.example.com> <871tbjpr4m.fsf@bsb.me.uk> <lnbnan8uf7.fsf@kst-u.example.com> <81b2f3fc-c570-4daf-850b-237640f1fd22@googlegroups.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Injection-Info: mx02.eternal-september.org; posting-host="945944de09706c9b4e29b53c9d2efdc2"; logging-data="32152"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1+gxx2csPXqpAWyLxmc62JX"
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/23.1 (gnu/linux)
Cancel-Lock: sha1:ZIYR4wW9vNXEh0ZF2vZziGyQLa0= sha1:I8FuoTzBziKKzOWAouRF2ryKhTc=
Xref: csiph.com comp.lang.c:76798

supercat@casperkitty.com writes:
> On Saturday, November 21, 2015 at 5:27:02 AM UTC-6, Keith Thompson wrote:
>> 2. sizeof ( vla_type )
>>    Not a constant expression.  Any expressions that are part of the type
>>    name are evaluated
>
> What if one has a VLA of pointers to a VLA type?  Would expressions forming
> part of the name of the thing being pointed to be evaluated?

Good question.  A concrete example:

    #include <stdio.h>

    int five(void) {
        puts("five()");
        return 5;
    }

    int six(void) {
        puts("six()");
        return 6;
    }

    int main(void) {
        printf("%zu\n", sizeof (char(*[five()])[six()]));
    }

On my system (where pointers are 8 bytes), the result of the
sizeof expression is 40, and both five() and six() are called.
The call to six() is not necessary to determine the result of the
sizeof expression.

(I'm not sure whether it's possible for the size to vary depending
on the value returned by six().)

I suggest that all expressions that are part of the type name should
be evaluated, just for the sake of keeping the description simple.
The standard could, I suppose, specify which expressions are needed
to determine the size and which are not, but it would be complex,
potentially ambiguous, and not worth the effort.  I could live
with it being unspecified if that would make things easier for
some compilers.

If I wanted to do something similar to the above in real-world
code, I would almost certainly define a typedef, so the issue
wouldn't arise:

    typedef char(*funky_array[five()])[six()];
    printf("%zu\n", sizeof (funky_array));

-- 
Keith Thompson (The_Other_Keith) kst-u@mib.org  <http://www.ghoti.net/~kst>
Working, but not speaking, for JetHead Development, Inc.
"We must do something.  This is something.  Therefore, we must do this."
    -- Antony Jay and Jonathan Lynn, "Yes Minister"