bug in Win32LCPTranscoder::transcode ?

[jiang lei]

Hi folks,

I noticed that there was a change in Win32LCPTranscoder between 2_1_0 and
2_3_0.

The old version uses mbstowcs() to calculate the buffer length while the new
version is using a new function called calcRequiredSize(const char* const
srcText). Inside calcRequiredSize(), mblen() is being used to get max
character size:

unsigned int Win32LCPTranscoder::calcRequiredSize(const char* const srcText)
{
    if (!srcText)
        return 0;

    unsigned charLen = ::mblen(srcText, MB_CUR_MAX);
    if (charLen == -1)
        return 0;
    else if (charLen != 0)
        charLen = strlen(srcText)/charLen;

    if (charLen == -1)
        return 0;
    return charLen;
}

I found that  if the source string starts with multibyte character,
"transcode" in 2_3_0 will always fail. I tried calling setlocale(LC_ALL, "")
before transcode. It did change MB_CUR_MAX from 1 to 2. But mblen continued
to return -1 despite of my effort. The 2_1_0 version of transcode uses
mbstowcs() and it's working fine. Is that a bug of 2_3_0 or am I missing
something?

My environment is: Win2K Professional/VC++.Net2002 Standard (Both are
Japanese Edition)

Thanks for your attention.

regards

Lei




---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]