Re: Unicode support

[Markus Plail]

Hi Pawel, Hu!

* Pawel Kot writes:
>On Fri, 27 Sep 2002, Hu Gang wrote:
>>good propose. It look very cool.
>>
>>Here is the patch.

>The same note:
>+               case 2: /* the value is multi byte
>+                        * so can check the before character */
>+                       if ( *(dest - 1) == '?' ) {
>+                               *(dest - 1) = *dest;
>+                               *(dest)     = *(dest + 1);
>+                               return 1;
>+                       }

>I think I don't get it. Why do we even touch *(dest-1)? It may be outside
>our process address space. My understanding of the wctomb is that we give
>it some encoded wide character. And we get the string that may be 1
>character long, 2 characters long, up to MB_CUR_MAX characters long.
>So we store it in the dest. And move the pointer the length - 1 bytes, as
>the loop iteration will move it one byte further.

>So the question: why do we even bother with '2' value? I think we don't
>need to do this at all. But I just woke up, care to enlighten me what do
>I miss?

I thought the same. And with my tests with Hu's test app wctomb always
returned 1 not 2.

So I throw the following code into the ring ;-)

void char_decode_unicode(unsigned char* dest, const unsigned char* src, int len)
{
        int i;

        for (i = 0; i < len; i++)
                wctomb(dest + strlen(dest), (src[i * 2] << 8) | src[(i * 2) + 
1]);
        return;
}

Works with both my 6210 folders and Hu's test app. 

regards
Markus