Path: ...!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: Keith Thompson <Keith.S.Thompson+u@gmail.com>
Newsgroups: comp.lang.c
Subject: Re: relearning C: why does an in-place change to a char* segfault?
Date: Tue, 13 Aug 2024 13:08:16 -0700
Organization: None to speak of
Lines: 56
Message-ID: <8734n8gp8v.fsf@nosuchdomain.example.com>
References: <IoGcndcJ1Zm83zb7nZ2dnZfqnPWdnZ2d@brightview.co.uk>
	<v8fhhl$232oi$1@dont-email.me> <v8fn2u$243nb$1@dont-email.me>
	<87jzh0gdru.fsf@nosuchdomain.example.com> <865xs54fak.fsf@linuxsc.com>
	<v9fqtb$3t7ph$4@dont-email.me>
MIME-Version: 1.0
Content-Type: text/plain
Injection-Date: Tue, 13 Aug 2024 22:08:17 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="75b586a96d8f13b1a18286173fed3ce1";
	logging-data="43115"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX18cxdo2B29msP9TgKcmg8ZF"
User-Agent: Gnus/5.13 (Gnus v5.13)
Cancel-Lock: sha1:0mpwo5p9qSP3nJm79LQD11QeKi4=
	sha1:n5zQ4JyCG0nr5Y8vrfZU35kT1tQ=
Bytes: 3042

Vir Campestris <vir.campestris@invalid.invalid> writes:
> On 12/08/2024 22:11, Tim Rentsch wrote:
>> Keith Thompson <Keith.S.Thompson+u@gmail.com> writes:
>> [...]
>> 
>>> A string literal creates an array object with static storage
>>> duration.  [...]
>> A small quibble.  Every string literal does sit in an array,
>> but it might not be a _new_ array, because different string
>> literals are allowed to overlap as long as the bytes in the
>> overlapping arrays have the right values.
>
> And this is exactly why string literals should always have been const.
>
> A compiler is entitled to share memory between strings. so
>
> 	puts("lap");
> 	puts("overlap");
>
> it's entitled to make them overlap. Then add
>
> 	char * p = "lap";
> 	*p='X';
>
> and it can overwrite the shared string. I think. which would mean that
> writing "lap" again would have a different result.
>
> But that ship has sailed. I'm not even sure const had been invented
> that far back!

The reason that *wasn't* done is that it would have broken existing
code.

In pre-ANSI C, "const" didn't exist, and this:
    char *ptr = "hello";
was the only way to create a pointer into a string literal object.
Making string literals const would have broken such code, with no clean
way to rewrite it so it would be accepted by old and new compilers.

I suppose you could do something like:

#ifndef __STDC__
#define const
#endif

In 20/20 hindsight, my personal opinion is that it would have been
better to make string literals const in C89/C90.  Compilers could
still accept old const-incorrect code with a non-fatal warning,
and programmers would be encouraged but not immediately forced to
use const.

This could still be done in C2y, but I'm not aware of any proposals.

-- 
Keith Thompson (The_Other_Keith) Keith.S.Thompson+u@gmail.com
void Void(void) { Void(); } /* The recursive call of the void */