DBCS/MBCS support

[spstream]

Hi Swen - Kudos for building this :)

It would be great to add support for:

Mixed SBCS + DBCS/MBCS text rendering in the same screen and field.

Correct input, display, and editing of double-byte and multibyte characters.

Proper cursor movement and character width handling for DBCS/MBCS text.

Code-page and locale-aware session settings.

Copy/paste and search behavior that preserves multibyte characters.

Validation against national-language host environments.

[el-dockerr]

Thank you for the kudos and for putting this on my radar! I completely agree that proper DBCS/MBCS support would be a massive addition to DX3270 (X3270 will be renamed to avoid confusions with the already existing x3270), especially for international mainframe environments.

To be completely candid, while I absolutely plan to implement this, it is going to take a considerable amount of time. I built the initial version of this emulator over a weekend, but adding double-byte support is not a quick patch—it requires a fundamental architectural shift. I am putting this on the roadmap, but it will be scheduled for a later release.

Here is a transparent look at what needs to be overhauled under the hood to make this work correctly:

• Data Stream Parsing: The engine needs a new state machine to handle IBM's Shift-Out (0x0E) and Shift-In (0x0F) control bytes, dynamically switching the parser between single-byte and double-byte reading modes.

• Screen Buffer Refactoring: Currently, the std::vector architecture assumes one byte equals one terminal cell. DBCS requires tracking "Left Half" and "Right Half" cells, ensuring that typing over one half automatically invalidates the other, and updating cursor navigation to jump two columns at a time.

• CoreText Rendering: The bundled IBM 3270 font does not contain CJK (Chinese, Japanese, Korean) glyphs. I will need to build a font-fallback mechanism to macOS system fonts while forcing those native glyphs to strictly adhere to our fixed-width, two-cell grid metrics.

• macOS Input (IME): To allow proper typing, the TerminalView must implement NSTextInputClient so the native macOS Input Method Editor (the floating OS window for Pinyin/Romaji) works seamlessly, followed by translating the resulting UTF-8 input back into EBCDIC DBCS.

The biggest hurdle for me as an independent developer is that public mainframes are generally strictly SBCS (like CP037). I do not currently have access to a live DBCS-enabled host to test against.

I am leaving this issue open and tagged as a long-term enhancement. I will post updates here once the buffer refactoring begins. Thanks again for the feature request!