Plugin search and Install instructions

[doctorfree]

Hi, I really like this service. Good job!

I have a couple of issues I would like to report.

First, the search for plugins used by a configuration seems to pickup references to plugins in comments.

Second, the installation instructions includes instructions on setting NVIM_APPNAME but my latest config got instructions to use a path:

git clone git@github.com:doctorfree/CandyVim ~/.config/doctorfree/CandyVim
NVIM_APPNAME=doctorfree/CandyVim/ nvim --headless +"Lazy! sync" +qa

Cloning a Neovim configuration into a subdirectory in ~/.config/<subdir>/ kinda works in 0.9 but has been disabled in Neovim nightly, if NVIM_APPNAME is set to a path Neovim nightly will issue an error message and exit.

So these instructions will eventually fail and even now when it appears to work, it breaks tempname() so any plugin using that will fail.

I've opened a discussion in the Neovim repository but so far there does not appear to be much interest in supporting this feature.

Just FYI, thanks for the excellent work.

[codicocodes]

Thank you for the comment! Sorry for the slow response I have not following up on the discussions tab!

I realized I might have had to address the install instructions since they had gotten stale on nightly. Fortunately though, it looks like Rory was able to merge a fix into nightly so it should probably be fine now - I believe!

neovim/neovim#25233

[doctorfree]

Yes, relative paths in NVIM_APPNAME is now supported in nightly. Thanks for helping with that!

I think tempname() is still broken for subdirs in NVIM_APPNAME in 0.9.x until that fix is backported from nightly. But, I never noticed a problem so maybe it's rare for plugins to need that? At any rate, I'm glad we can use a subdirectory for configs now.

[codicocodes]

Regarding the plugin search picking up commented plugins,T I am doing quite simple checking for plugins based on just strings comparison and regex. I think it works well for 99% of use cases, and I am currently not convinced it would make sense to do more advanced parsing to strip out comments first.

Would be open to a proposal/contributions on it, but still quite concerned it would increase the runtime of parsing all files in all configs (currently 800+ configs)