Encoding

Status: Standard (ISO 2022 character-set switching; UTF-8 per RFC 3629).

The byte stream on a PTY has no declared encoding. The emulator decodes it using a mode selected by configuration, locale, or an escape sequence, and the application must agree without being told.

UTF-8 as the practical default

Every emulator in the compatibility tables defaults to UTF-8 when the locale says so, and most do so unconditionally. LANG or LC_CTYPE ending in .UTF-8 is the signal applications and libraries such as ncurses use; the emulator often sets nothing and simply decodes UTF-8.

The explicit switches are:

ESC % G    select UTF-8
ESC % @    return to ISO 2022 / Latin-1 handling

xterm honors both. Several modern emulators ignore them and stay in UTF-8; they are safe to send but cannot be relied on to change anything.

Legacy character sets

ISO 2022 lets a terminal hold up to four designated sets, G0 to G3, and shift between them:

Sequence	Meaning
ESC ( B	Designate US-ASCII into G0
ESC ( 0	Designate DEC Special Graphics (line drawing) into G0
ESC ) B, ESC ) 0	Same, into G1
ESC * …, ESC + …	Same, into G2 and G3
SO (0x0e)	Lock shift: use G1
SI (0x0f)	Lock shift: use G0
ESC N, ESC O	Single shift: next character from G2 or G3

DEC Special Graphics is the one legacy set still in daily use. Terminfo's acsc capability maps ASCII letters to line-drawing glyphs through it, and ncurses emits ESC ( 0 followed by j, k, l, m, q, x for corners and lines. Most UTF-8 emulators still translate this set even when they implement nothing else from ISO 2022.

luit is xterm's companion filter that converts legacy locales to UTF-8 on the fly. It remains the reference for how a UTF-8 emulator can serve a non-UTF-8 application.

Invalid input

A decoder must not stall on malformed bytes. The common choices are to emit U+FFFD for each invalid byte or for each maximal invalid subsequence, as described in the Unicode Standard chapter 3. Emulators differ in how many replacement characters one bad sequence produces, which changes the cursor position; applications should never write bytes they do not know to be valid.

C1 controls

Bytes 0x80–0x9f are C1 control codes in ISO 6429 but continuation bytes in UTF-8. In a UTF-8 session an emulator must treat them as text, or a valid character such as U+2013 would be parsed as CSI. xterm documents that it disables 8-bit controls in UTF-8 mode; modern emulators do not recognize them at all. Always send the two-byte ESC [ forms.

Byte-order mark

U+FEFF at the start of a stream is not meaningful on a PTY. Emulators treat it as a zero-width character; applications should strip it from files before printing.

Probe

printf '\033(0lqqk\033(B\n'      # expect a box corner: ┌──┐
printf '\xe2\x80\x93\n'          # en dash; must print, not swallow the line
printf '\xff\n'                  # one invalid byte; count the U+FFFD glyphs
printf '\033%%G'                 # harmless in UTF-8; no visible change

Sources

• ISO/IEC 2022
• RFC 3629, UTF-8
• XTerm Control Sequences
• luit
• Unicode Standard, chapter 3, conformance

Compatibility

Feature support by terminal. Imported claims have not been reverified; unknown is not unsupported.

Feature / stable ID	Alacritty	Apple Terminal	ConPTY (conhost)	Contour	DomTerm	foot	Ghostty	iTerm2	kitty	Konsole	mintty	mlterm	PuTTY	Revenant	RLogin	st	tmux	rxvt-unicode	VTE	wayst	WezTerm	Windows Terminal	xterm	xterm.js
ESC % G / ESC % @ esc-percent-encoding	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	SupportedImported · unverified	Unknown
DEC Special Graphics esc-dec-special-graphics	SupportedImported · unverified	Unknown	Unknown	Unknown	Unknown	SupportedImported · unverified	SupportedImported · unverified	Unknown	SupportedImported · unverified	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	SupportedImported · unverified	Unknown	SupportedImported · unverified	SupportedImported · unverified	SupportedImported · unverified	SupportedImported · unverified
8-bit C1 in UTF-8 text-c1-in-utf8	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	UnsupportedImported · unverified	Unknown
Non-UTF-8 locales text-non-utf8-locales	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	UnsupportedImported · unverified	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	Unknown	SupportedImported · unverified	Unknown