From:	"Tim Nicholson" <T DOT J DOT Nicholson AT btinternet DOT com>
Newsgroups:	comp.os.msdos.djgpp
Subject:	Re: posible bug in GCC syntax checker?
Date:	Fri, 19 Oct 2001 19:01:24 +0100
Organization:	Skyforce avionics Limited
Lines:	59
Message-ID:	<9qppug$9pv$1@uranium.btinternet.com>
References:	<9q4nm9$lc9$1 AT plutonium DOT btinternet DOT com> <Xns9137B8BB094D0ASINANUNUR AT 132 DOT 236 DOT 56 DOT 8>
NNTP-Posting-Host:	host213-1-60-131.btinternet.com
X-Priority:	3
X-MSMail-Priority:	Normal
X-Newsreader:	Microsoft Outlook Express 5.50.4133.2400
X-MimeOLE:	Produced By Microsoft MimeOLE V5.50.4133.2400
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp
Reply-To:	djgpp AT delorie DOT com

Yes, I had already discovered that it worked without the const. I always
like to use const when passing a pointer argument to a function that is not
designed to change it. That way I get a warning if, for whatever reason, the
function or any other function down the line tries to write to the address.
After all that is what const is for isn't it?

I have visited a number of newsgroups with regard to this and It would seem
that this is not a bug but an ANSI feature! I find it a real pain as it
removes some of the flexibility that made C such a powerful programming
tool.

I do not use C++ in DJGPP but I do use Borland C++ Builder and VC++ when I
have to :-( As you say, they do not exhibit this behaviour. I have heard
some very reasoned arguments both for and against and I still remain
unconvinced that this was really intentional. ANSI is meant to be a
standard, but like many standards it does seem to be open to different
interpretations and implementations. I have to write code that does not meet
our company's most basic coding standards (i.e no const where ther should be
one) in order to please the compiler - I doubt that is what the committee
intended.

I guess this is now a little of topic for here but, at the end of the day we
all use DJGPP to compile our code. It was that implementation of the GCC
compiler that produced the unexpected warning - so I do feel there is
validity in the origional post at least. Now prehaps this topic should be
left to the .lang. boys and girls :-)



Tim




"A. Sinan Unur" <asu1 AT cornell DOT edu> wrote in message
news:Xns9137B8BB094D0ASINANUNUR AT 132 DOT 236 DOT 56 DOT 8...
> "Tim Nicholson" <T DOT J DOT Nicholson AT btinternet DOT com> wrote in
> news:9q4nm9$lc9$1 AT plutonium DOT btinternet DOT com:
>
> > Warning: Passing arg 1 of draw_item from incompatible pointer type
>
> It probably isn't a bug with GCC although I am no expert in rules
regarding
> the decay of multi-dimensional arrays to pointers. Your code compiles
> without warnings if const is removed.
>
> Note that the code also compiles without warnings if compiled as C++.
>
> The answer to your question is tied to these facts but I do not know what
> the answer is. You may want to check with the folks on
> comp.lang.c.moderated.
>
> sinan.
> --
> --------------------------------
> A. Sinan Unur
> http://www.unur.com/

- Raw text -