Path: csiph.com!news.mixmin.net!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: Effect of CPP tags
Date: Mon, 08 Jan 2024 10:32:07 -0800
Organization: None to speak of
Lines: 33
Message-ID: <871qarr80o.fsf@nosuchdomain.example.com>
References: <umet9d$3hir9$1@dont-email.me> <RJAlN.141573$c3Ea.132913@fx10.iad> <un6toq$3ogqm$1@dont-email.me> <%cFlN.140487$xHn7.115393@fx14.iad> <slrnupeb0t.qme.jj@iridium.wf32df> <un7cjh$3qi9d$1@dont-email.me> <un7eoj$3qpmp$1@dont-email.me> <un7ndt$3rr75$1@dont-email.me> <un92c7$6325$1@dont-email.me> <un9va1$a752$1@dont-email.me> <unb5d3$hum9$2@dont-email.me> <unb9uk$igst$1@dont-email.me> <87jzomrvh8.fsf@nosuchdomain.example.com> <uncq37$pvae$1@dont-email.me> <878r52rljn.fsf@nosuchdomain.example.com> <und22o$qs4c$1@dont-email.me> <874jfpstyc.fsf@nosuchdomain.example.com> <unefqu$148mk$1@dont-email.me> <87sf38r9bu.fsf@nosuchdomain.example.com> <unfjb3$195hu$1@dont-email.me> <unh2nb$1iplq$1@dont-email.me>
MIME-Version: 1.0
Content-Type: text/plain
Injection-Info: dont-email.me; posting-host="f909ce6842d964f6deb0fc108ae5e13b"; logging-data="1723433"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX19MLJjM3OI2w3kCrWy+1yYp"
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/27.2 (gnu/linux)
Cancel-Lock: sha1:4I1qH4t1AF1IaE3+bUCY9uzjp4U= sha1:mb3KpmXE4JgXsyR98iEkomKe2Wc=
Xref: csiph.com comp.lang.c:379930

David Brown <david.brown@hesbynett.no> writes:
> On 08/01/2024 02:32, Bart wrote:
[...]
>> (BTW why did VLAs become optional from C11?)
>
> I don't know for sure.  But I do know that not all C implementations
> use a stack, and for some targets a "true VLA" (as distinct from a VLA
> where the size is known at compile time) would be extremely
> inefficient to implement.  It is possible that this has something to
> do with it.

I'm skeptical that that's the reason.  Almost all C implementations do
use a "stack", in the sense of a contiguously allocated region of memory
in which automatic objects are allocated, with addresses uniformly
increasing or decreasing for new allocations.  (All C implementations
use a "stack" in the sense of a last-in/first-out data structure.)  I'm
not aware that any implementers of non-stack implementations objected to
VLAs.  For that matter, I don't know why VLAs would be extremely
inefficient in such an implementation.  They need to have the ability to
allocate new stack frames anyway, and determining the allocation size at
run time shouldn't be a huge burden.

I've seen suggestions that the intent was to make it possible to create
conforming C implementations for small embedded processors.  That might
make sense, but it could have been addressed by making VLAs optional
only for freestanding implementations.

I think Microsoft has chosen not to implement VLAs in their C compiler.

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