Regression in prompt resetting

[ajvondrak]

General information

• Terminal program: iTerm2 Build 3.4.19
• Operating system: macOS Ventura 13.3.1 (M1 macbook running iTerm2 in Rosetta mode)
• ZSH framework: zgenom @ 2b55d9d9
• ZSH version: zsh 5.9 (x86_64-apple-darwin22.1.0)
• ZVM version: c4fc3a60d0c8ca9f435dd0e28cdc6ec42c4b7ad8

Basic examination

• [x] I have read through the README page
• [x] I have the latest version of zsh-vi-mode
• [ ] I have tested with another terminal program

Problem description

I have a custom PROMPT that relies on prompt_subst to change based on whether I'm in Vi mode. However, upon upgrading zsh-vi-mode, this prompt stopped switching at all.

Reproduction steps

1. Configure a custom prompt:

   setopt prompt_subst
   PROMPT='$(custom_prompt)'
   custom_prompt() {
     if [[ $KEYMAP == "vicmd" ]]; then
       echo -n "vi cmd mode: "
     else
       echo -n "insert mode: "
     fi
   }

2. Hit <esc>/i to toggle between insert mode & vi cmd mode.
   • Expect: prompt to change between vi cmd mode: and insert mode:
   • Actual: prompt stays insert mode:
3. Upon reverting 5979662a11e139a3a87872d1a5282c836e98e608 (and doing a zgenom reset and everything to make sure I'm running an up-to-date zsh-vi-mode plugin), the above test toggles the prompt properly.

Expected behavior

I'm not sure what issue #180 was fixing (I don't use tmux), but the fix that got merged appears to cause this regression. It looks related to the prompt-resetting logic changes.

[ajvondrak]

Might be related to / a duplicate of #213?

[jeffreytse]

Hi @ajvondrak

Thanks for your reporting. I still need some time to investigate this issue, please be patient for the update.

Thanks & Regards

[ajvondrak]

Understood & much appreciated. 🙏 In the meantime, I can work locally by just reverting the commit in question, so I'm not really hurting too much. Thanks for all the great work on this plugin!

Message ID: @.***>

[jeffreytse]

Hi @ajvondrak,

Could you please check again with the latest commit? See if your issue has been settled down. : )

Thanks & Regards

[ajvondrak]

Just tried against a3d717831c1864de8eabf20b946d66afc67e6695, and my prompt works! 🎉

And to double-check, I went ahead and did a git checkout HEAD^ on zsh-vi-mode a few times by hand until the prompt stopped working. Findings:

• c8ca1a1bb69e3a116c1c348d0693f849de5f645a works
• fb2f825bf01012786b9a327e5e3c2085d8d18ae5 works
• 1f28e1886dc8e49f41b817634d5c7695b6abb145 does not work

Indicating that fb2f825bf01012786b9a327e5e3c2085d8d18ae5 was indeed the first commit to fix my problem (as noted, #243 tracked the fix to this issue as well). 👍

Thanks so much for the work here!