add wezterm detection/differentiation

[dankamongmen]

The wezterm FAQ explicitly disclaims any independent terminfo database, stating that xterm-256color ought be used. OK, that's fine. For advanced features beyond terminfo, however (see e.g. #1715 and #1758), we need be able to differentiate wezterm from the forty thousand other terminals claiming to be xterm.

I don't see any way to interrogate wezterm, but according to wez/wezterm#230 we can look for the `TERM_PROGRAM environment variable, and indeed:

declare -x TERM_PROGRAM="WezTerm"
declare -x TERM_PROGRAM_VERSION="20210610-204057-2e7dc70e"

alright, i guess we're using that.

[wez]

The versioning scheme is based on the date of the commit + the commit hash, so you could check to see if you got 20210610-204057 or later.

In addition to the environment, CSI > q should cause wezterm to answerback with the terminal name, a space and then the version string: DCS >| WezTerm VERS

[wez]

re: terminfo, we do have a wezterm terminfo file, but it isn't installed by default and most users won't have enabled it.
https://wezfurlong.org/wezterm/faq.html?highlight=terminfo#how-do-i-enable-undercurl-curly-underlines has details on how to install and use it, but the other methods of detection are more definitive.

[dankamongmen]

WezTerm is now detected based off the TERM_PROGRAM environment variable. it must be WezTerm, and TERM must contain the string xterm.

though now I see that you have a DCS-based method, in which case that will trump everything, once i add it (that work is ongoing, and intended to land this weekend).

[wez]

FWIW, wezterm can also be running with TERM=wezterm for some users, I'd suggest removing the TERM constraint when inspecting the environment.

If you're into answerback stuff, I commented above with the sequence and answerback that also returns the wezterm version info.

[wez]

(ah I can see that you saw my comment as well--great!)

[dankamongmen]

If you're into answerback stuff, I commented above with the sequence and answerback that also returns the wezterm version info.

yep, that's what'll be going in; i've got a branch rihgt now for DCS-based identification. and actually, as a result of that, i'm seeing:

 2021-06-12T17:12:19.351Z WARN  wezterm_term::terminalstate    > unknown unspecified CSI: "=0c"
 2021-06-12T17:12:19.351Z WARN  mux::localpane                 > unknown DeviceControlMode::Enter EnterDeviceControlMode(params: [], intermediates: ['+' 0x2b, ], byte: 'q' 0x71, ignored_extra_intermediates=false)
 2021-06-12T17:12:19.351Z WARN  mux::localpane                 > unhandled DeviceControlMode::Data 35 5
 2021-06-12T17:12:19.351Z WARN  mux::localpane                 > unhandled DeviceControlMode::Data 34 4
 2021-06-12T17:12:19.351Z WARN  mux::localpane                 > unhandled DeviceControlMode::Data 34 4
 2021-06-12T17:12:19.351Z WARN  mux::localpane                 > unhandled DeviceControlMode::Data 65 e

i can go file a bug if you'd like with the respective queries etc

[wez]

Yes please file it!

[dankamongmen]

confirmed reply to CSI > q:

:>|WezTerm 20210610-204057-2e7dc70e^[\^[[?1;2S^[[?2;0;960;624S^[[?1;0;65536S^[[?65;4;6;18;22c

i'll integrate this into my great shitty state machine in the sky