Martin,
The only valid argument against the patch would be that (const char **)
causes warnings to other systems. If it does not, why not apply it to keep
everybody, including Solaris and OSX, happy?
It is not an alternative to change libiconv sources or configuration,
especially if it doesn't make a difference for kannel.
BR,
Nikos
----- Original Message -----
From: "Martin Conte Mac Donell" <[email protected]>
To: <[email protected]>
Sent: Monday, March 02, 2009 4:28 AM
Subject: Re: gwlib/charset patch
On Sun, Mar 1, 2009 at 11:47 PM, Nikos Balkanas <[email protected]>
wrote:
Hi,
Nope. This is the function prototype from include/iconv.h:
extern LIBICONV_DLL_EXPORTED size_t iconv (iconv_t cd, const char* *
inbuf,
size_t *inbytesleft, char* * outbuf, size_t *outbytesleft);
Note the 2nd argument, const char* * inbuf. This is what makes gcc 3.4.3
(stock solaris 10) to complain. This is from iconv sources 1.12, general
not
solaris or linux specific and outside any OS related preprocesor
definitions.
Could be that newer gcc doesn't complain about (const char) -> (char)
conversions, but even if it doesn't, it doesn't hurt to be compatible to
the
function declaration, and make Solaris and OSX folks happy. :-)
BR,
Nikos
(OPS: Again but this time "replying all")
Nope, Alex is right.
Actually you are not looking at the sources, you are looking at the
generated libraries. libiconv comes with a iconv.h.in in which the
prototype is this:
extern size_t iconv (iconv_t cd, @ICONV_CONST@ char* * inbuf, size_t
*inbytesleft, char* * outbuf, size_t *outbytesleft);
So, the cast should be more like
#ifdef __APPLE__
#define ICONV_CONST const
#elif OPENBSD
#define ICONV_CONST const
#else
#define ICONV_CONST
#endif
(...)
ret = iconv(cd, (ICONV_CONST char**)&from_buf, &inbytes, &pointer,
&outbytes);
M.
