Tcl/Tk 9.1.0 shipped September 29, 2026, and the feature list reads like a quiet refutation of every "Tcl is dead" post from the last fifteen years. This is not a resurrection — it is a language that never stopped being maintained, now pulling itself closer to modern expectations in ways that matter. The Tcl-side highlights center on Unicode normalization (the new unicode command and corresponding C routines), which closes a gap that has been painful for anyone building internationalized applications. The timer command finally exposes a monotonic clock with microsecond resolution — a feature most modern languages ship with out of the box, and whose absence in Tcl forced workarounds that ranged from annoying to dangerous. lfilter adds a list-selection primitive that functional-programming users will recognize as overdue. Tk 9.1 is where the real ambition lives. Accessibility screen reader support and initial bidirectional text / RTL language handling are not cosmetic additions — they are table-stakes requirements for any GUI toolkit that wants to be used in production in 2026. The new ttk::toggleswitch widget and rotated text on labels modernize the visual vocabulary. Removing Windows XP appearance support is the kind of cleanup that signals the project has stopped trying to be everything to everyone and started deciding what it actually is. Under the hood, the changes are workmanlike and welcome: improved list internals for memory efficiency on large lists, extended 64-bit size support, case-insensitive filesystem paths on macOS, and a revised autoexecok/exec search on Windows. Applications are now required to call an initialization routine (TclFindExecutable or TclZipfsAppHook), which tightens the embedding contract. The C API gets several new routines — TclIsEmpty, TclGetEncodingNameForUser, TclAttemptCreateHashEntry, and list manipulation functions — plus a new time API using long long instead of TclTime. These are the kinds of changes that matter to embedders and extension authors, the people who keep Tcl alive in industrial automation, EDA tools, and network testing. None of this will convert skeptics or trigger a Tcl renaissance. What it does is keep the language viable for the communities that depend on it, and push Tk from "legacy toolkit" toward "toolkit with a modern conscience." The accessibility and RTL work in particular suggests someone on the core team is thinking about who actually uses this software, not just who used to.
