When zsh is run with -o vi / -o emacs, instead of loading ZLE to force
the change, only record the option state. Then when ZLE starts, the vi
option determines the default keymap.
They are likely to be terminal query responses not keys. If they do match a
keybinding that's not affected. Also does likewise for certain CSI sequences.
Python modules can behave like commands in their own right. This allows
modules to define _python_module-* functions that are used to complete
module arguments. Also gets the ball rolling by defining completions for
venv, http.server, and json.tool.
1) As for now, `--local-user` completion is irrelevant: it completes system users, while gpg expects private keys IDs
2) `secret-keys` completion was also broken and completed nothing.
The reason of that to happen was in that fact that it assumed `uid` token would be next after `fpr` (same as it do in public-keys completion), while in current gnupg versions there is at least `grp` token, and potentially can be others.
So, instead i+=2 I made `until "uid"` loop.
update the following features to use the monotonic clock for calculating time
deltas and intervals:
* MAILCHECK parameter
* PERIOD parameter
* SECONDS parameter
* %(nS.t.f) prompt-expansion sequence
* time built-in's elapsed time and cpu % values
* zsh/zftp ZFTP_TMOUT parameter
* zsh/zprof timings
also use CLOCK_MONOTONIC_RAW instead of CLOCK_MONOTONIC on macOS
- fix broken page completion on macOS and FreeBSD (regression caused by
workers/50278)
- update fall-back paths to include /usr/local/share/man (fixes issues like
workers/53249 for systems without `manpath` or equivalent)
- fold macOS completion into FreeBSD's, as macOS 13+ uses it now with little
modification
- add -K option for FreeBSD and -h option for mandoc/NetBSD, complete
pre-processor sequences, complete section lists with commas for man-db
- exclude HTML pages for FreeBSD-likes
