GUIX Multi-line Text View Word Wrapping Broken for UTF-8 Strings

[ThomasBurgess2000]

Describe the bug
The GUIX multi-line text view widget has a bug in its word wrapping logic. The word boundary detection fails inconsistently, causing words to be split mid-character instead of at proper word boundaries (spaces, commas, semicolons).

I'm using GUIX 6.1.11

To Reproduce

1. Create a multi-line text view widget
2. Set text containing spaces, such as: "Hola. Esta es una demostractión de inglés"
3. Observe inconsistent word wrapping behavior

It appears that breaks work properly at punctuation like commas but not spaces? I think this is a UTF-8 issue

Inside the loop:

ch = string;

#ifdef GX_UTF8_SUPPORT
    _gx_utility_utf8_string_character_get(&string, GX_NULL, &glyph_len);
    current_index += glyph_len;
#else
    string.gx_string_ptr++;
    string.gx_string_length--;
#endif /* GX_UTF8_SUPPORT */

ch.gx_string_length = glyph_len;

So ch is meant to represent the current character (possibly multi-byte). But immediately afterwards, the code does single-byte checks like:

if (ch.gx_string_ptr[0] == GX_KEY_CARRIAGE_RETURN)
...
else if (ch.gx_string_ptr[0] == GX_KEY_LINE_FEED)
...
else if (((text_info -> gx_text_display_width + char_width) > available_width - 1) &&
         (text_info -> gx_text_display_number > 0) &&
         (ch.gx_string_ptr[0] != ' '))
...
if ((ch.gx_string_ptr[0] == ' ') || (ch.gx_string_ptr[0] == ',') || (ch.gx_string_ptr[0] == ';'))

For ASCII, ch.gx_string_ptr[0] works fine, since one character == one byte. But for UTF-8, ch.gx_string_length may be >1, but the code only checks the first byte of the UTF-8 sequence. Non-ASCII spaces (e.g. U+00A0 non-breaking space, U+3000 ideographic space) will never be recognized as valid breakpoints, because their first UTF-8 byte isn’t ' ' (0x20).

Also, this condition refuses to backtrack when the overflowing glyph is a space:

else if (((text_info->gx_text_display_width + char_width) > available_width - 1) &&
         (text_info->gx_text_display_number > 0) &&
         (ch.gx_string_ptr[0] != ' '))
{
    if (display_number == 0) {
        break;
    }
    text_info->gx_text_display_width = display_width;
    text_info->gx_text_display_number = display_number;
    break;
}

Expected behavior

• Words should break at natural boundaries (spaces, punctuation)
• Long words that exceed line width should break at word boundaries when possible
• Consistent behavior between ASCII and UTF-8 text

Impact
Annoyance, bad experience

[fdesbiens]

Thank you for raising this issue, @ThomasBurgess2000.

I will bring it to the team's attention.

[fdesbiens]

We discussed the problem. Unfortunately, there is no easy fix for it.

Our team is small and the people who have been working on GUIX at Express Logic and Microsoft are no longer involved.

I will move this issue to our backlog, as it flags a genuine problem in the tool. However, please do not expect a quick resolution. I am the only team member with some knowledge of the MFC framework, and I have many competing priorities.