visual c++ .net 2003 bug in c4090 waring

[Murali Yeddanapudi]

Hi

The C code (not C++) below (test.c) generates a warning:

#include <stdlib.h>
int main()
{
const char** a = (const char**) malloc(1*sizeof(const char*));
a[0] = "boo";
free(a);
return(0);
}

If I compile above with MSVC 7.1:

I:\tmp>cl test.c
Microsoft (R) 32-bit C/C++ Optimizing Compiler Version 13.10.3077 for 80x86
Copyright (C) Microsoft Corporation 1984-2002. All rights reserved.

test.c
test.c(6) : warning C4090: 'function' : different 'const' qualifiers
Microsoft (R) Incremental Linker Version 7.10.3077
Copyright (C) Microsoft Corporation. All rights reserved.

/out:test.exe
test.obj

I believe this is not correct as I am not freeing the const data that is
being pointed to, only the non-const array of pointers.

Is this a compiler bug or am I missing sometihng?

Thanks,
Murali
--

[David Lowndes]

>The C code (not C++) below (test.c) generates a warning:

...
If I compile above with MSVC 7.1:

Murali,

It also does with VC2005 if compiled as 'C' code.

The Comeau online compiler doesn't object, so it may well be an issue
with the MS compiler. I suggest that you submit a bug report on it
against VC2005 (mention that VC2003 is the same too):

http://lab.msdn.microsoft.com/productfeedback/

Post back a link to your bug report and I'll confirm it.

Dave

[Murali Yeddanapudi]

Hi Dave:

Here is the bug report I created:

http://lab.msdn.microsoft.com/Produc...6-910a36a493ab

A similar one appears to have been reported already (and is marked
not reproducible, not sure why), albeit with regards to memset, but I
think the problem is the same.

http://lab.msdn.microsoft.com/produc...f-95ddb65117e9

Regards,
Murali
--

PS: On an unrelated note, I first tried reporting this as a bug by going to
microsoft.com. I was expecting to find a link where I could put in the
relavant
info and press submit. The best I could do was find this newgroup to post.
Later in the day a collegue pointed me to the lab.* link you mention, I am a
bit
surprised that a link to report bugs (free QE from my point of view) is not
easily found on from the support.microsoft.com web-ste.

-------
"David Lowndes" wrote:

The C code (not C++) below (test.c) generates a warning:
...
If I compile above with MSVC 7.1:

Murali,

It also does with VC2005 if compiled as 'C' code.

The Comeau online compiler doesn't object, so it may well be an issue
with the MS compiler. I suggest that you submit a bug report on it
against VC2005 (mention that VC2003 is the same too):

http://lab.msdn.microsoft.com/productfeedback/

Post back a link to your bug report and I'll confirm it.

Dave

[David Lowndes]

>Here is the bug report I created:

http://lab.msdn.microsoft.com/Produc...6-910a36a493ab
I've validated it and added a comment.
A similar one appears to have been reported already (and is marked
not reproducible, not sure why)...

Sometimes you have to persist with them. It helps when you have the
ability to re-open the bug reports to get them re-checked.

Cheers
Dave

[Paul Forgey]

I noticed MS closed it out as by design. I very strongly disagree and it's
sort of a pain for us at our organization because we use -Werr and
simultaneously use different compilers.

You might want to reopen the bug (please). The resopnse from Microsoft
demonstrates a misunderstanding of the const modifier and ANSI C. You are
correct that char ** is compatible with void *.

The warning would be valid if the paramter being passed was const char *
const *.


"Murali Yeddanapudi" wrote:

Hi Dave:

Here is the bug report I created:

http://lab.msdn.microsoft.com/Produc...6-910a36a493ab

A similar one appears to have been reported already (and is marked
not reproducible, not sure why), albeit with regards to memset, but I
think the problem is the same.

http://lab.msdn.microsoft.com/produc...f-95ddb65117e9

Regards,
Murali
--

PS: On an unrelated note, I first tried reporting this as a bug by going to
microsoft.com. I was expecting to find a link where I could put in the
relavant
info and press submit. The best I could do was find this newgroup to post.
Later in the day a collegue pointed me to the lab.* link you mention, I am a
bit
surprised that a link to report bugs (free QE from my point of view) is not
easily found on from the support.microsoft.com web-ste.

-------
"David Lowndes" wrote:

The C code (not C++) below (test.c) generates a warning:
...
If I compile above with MSVC 7.1:

Murali,

It also does with VC2005 if compiled as 'C' code.

The Comeau online compiler doesn't object, so it may well be an issue
with the MS compiler. I suggest that you submit a bug report on it
against VC2005 (mention that VC2003 is the same too):

http://lab.msdn.microsoft.com/productfeedback/

Post back a link to your bug report and I'll confirm it.

Dave

[David Lowndes]

>You might want to reopen the bug (please). The resopnse from Microsoft

demonstrates a misunderstanding of the const modifier and ANSI C. You are
correct that char ** is compatible with void *.

I've re-opened it - let's hope for a better response this time!

Dave

[Paul Forgey]

I wish I could quote chapter and verse of ANSI C-99 to which this applies.
If your organization happens to have a copy, try quoting it so you don't have
to defer to the authority of another compiler. Their C++ compiler does
properly handle this by the way.

I mis-typed in the earlier response. I meant to say const char ** is
compatible with void *, not char ** (which of course is also compatible).

In the meantime you can silence the warning with an explicit cast to (void
*), defeating the const violation checking.

"David Lowndes" wrote:

You might want to reopen the bug (please). The resopnse from Microsoft
demonstrates a misunderstanding of the const modifier and ANSI C. You are
correct that char ** is compatible with void *.

I've re-opened it - let's hope for a better response this time!

Dave