Path: csiph.com!fu-berlin.de!uni-berlin.de!individual.net!not-for-mail
From: Ian Collins <ian-news@hotmail.com>
Newsgroups: comp.lang.c
Subject: Re: usecases of void foo(char **b) {}
Date: Sat, 5 Nov 2016 16:34:32 +1300
Lines: 43
Message-ID: <e852a8Fpc20U4@mid.individual.net>
References: <f04d8120-e4c9-464b-923c-7dc262da5b5b@googlegroups.com> <nv8r1c$i54$1@dont-email.me> <87mvhjo3jw.fsf@bsb.me.uk> <nv9ubc$7uf$1@dont-email.me> <b4a396b5-48ff-485e-845b-dca9c66de64c@googlegroups.com> <lneg2v42t8.fsf@kst-u.example.com> <3a09f5ec-d4f7-40fa-9855-b5c3f02ce4e9@googlegroups.com> <87ziljks6k.fsf@bsb.me.uk> <2199a2dd-d844-4f8d-bfe0-6a7c2854277b@googlegroups.com> <87twbqlsin.fsf@bsb.me.uk> <nvbaii$arc$1@dont-email.me> <87oa1yljof.fsf@bsb.me.uk> <nvcdp8$rvt$1@dont-email.me> <87ins5k15t.fsf@bsb.me.uk> <nvdv04$g49$1@dont-email.me> <87bmxxm6zs.fsf@gmail.com> <nvf4ra$dk0$1@dont-email.me> <lnshr8zhx1.fsf@kst-u.example.com> <nvfor6$nv8$1@dont-email.me> <ln7f8kz8c6.fsf@kst-u.example.com> <nvgq64$ho3$1@dont-email.me> <871syrijhw.fsf@bsb.me.uk> <nvhtuu$jun$1@dont-email.me> <aeccd37e-50de-411d-a9cd-7c17c44d8204@googlegroups.com> <nvir4h$1c7$1@dont-email.me> <e84j63Fpc20U3@mid.individual.net> <nvjcq6$r1a$1@dont-email.me>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
X-Trace: individual.net qQR33f3Ruw8zuABedJ5SXQ8U0jZ9uGMcbEamOzUPVX6Wk+3djz
Cancel-Lock: sha1:DfLVO/+Q6cKgnjaJy9I9vu1oWqQ=
User-Agent: Mozilla/5.0 (X11; SunOS i86pc; rv:38.0) Gecko/20100101 Thunderbird/38.5.1
In-Reply-To: <nvjcq6$r1a$1@dont-email.me>
Xref: csiph.com comp.lang.c:95822

On 11/ 5/16 02:32 PM, BartC wrote:
> On 04/11/2016 23:16, Ian Collins wrote:
>> On 11/ 5/16 09:30 AM, BartC wrote:
>>> On 04/11/2016 19:55, asetofsymbols@gmail.com wrote:
>>>> Bart wrote:
>>>> Instead I not understand
>>>>
>>>>> typedef int (*Matrix)[columns];
>>>>
>>>> What I like is:
>>>> int *matrx=malloc (size);
>>>> All linear even if it is
>>>> one nxm matrix...
>>>
>>> Well, the matrix being passed here /is/ linear, as the original is
>>> int[][] so they must match.
>>>
>>> What's happening is the same as when you use int* when the original is
>>> int[].
>>>
>>> Matrix points to one row of the original, of width <columns>. Which is
>>> another puzzling thing as 'columns' is a variable. Did you even know
>>> that that C allows variables when declaring array dimensions, and in a
>>> typedef? This doesn't quite fit the pattern of VLA as the only thing
>>> allocated locally is a pointer. (And what happens if 'columns' was to
>>> change after Matrix is declared?)
>>>
>>> The obvious (to me) way of declaring Matrix would be as
>>> int(*)[][columns]. The original is int[][]; this is a pointer to that.
>>
>> If you find this stuff to complicated in C, why don't you use C++ where
>> the types you want either already exist, or can be built?
>
> What I'm talking about already exists in C. But people are using every
> excuse not to use it.

My point was if you don't like the ambiguity of arrays and pointers, in 
C++ if you want to pass an array, you pass an array type.  If you want 
to pass a matrix, you pass a matrix type.  In C++ you hardly ever have 
to pass a naked pointer.

-- 
Ian