The ollama plugin was using some APIs that are not yet available in nvim
stable. Now we are using the older APIs we can use this on other
versions of nvim.
The `vim.api.nvim_buf_set_text` does not take some parameters as `-1` to
add text to the end of the line. Now we are getting the text on the last
line adding the new text to it, then replacing the last line.
This is the plugin I have been using for a while now. I think its much
nicer to use than a tree style. Treating the filesystem as a buffer is
something I used in emacs and have come quite used to.
This is mainly for stylelint, I don't always have this installed and it
throws an error when you open a sass file and it tries to lint the
buffer. Now it will not know about any linters unless they are in the
PATH.
Eslint has become slow for some projects for reasons unknown. In an
attempt to speed up my linting process, I'm experimenting with eslint_d,
which keeps a process running. For the time being, I'd like to keep
linting and formatting separate from the Language Server Protocol (LSP)
to avoid the need to install every LSP. This way, I can have `shfmt`
without the bash LSP, for example.
For JavaScript, I may consider moving this into an LSP since I use it
for work, which would justify combining responsibilities. However, for
now, I'll be giving eslint_d a try.
Indent Line is no longer in use; it's quite an old plugin, and there are
likely better ways to implement this functionality. For now, I won't be
using any indent markers, and we'll see how that goes. If I encounter
issues, I will explore alternative plugins.
Orgmode was an attempt to bring over the knowledge base from Emacs. Vim
doesn't have the same workflow as Emacs, and everything is less tightly
integrated into a single application. I have now transitioned to using
Markdown and Obsidian, focusing on the more widely-used Markdown
integration.
Conjure is too heavy for my needs. All I require is a simple plugin to
execute Clojure in a REPL. This plugin conflicts with many other plugins
and filetypes, where the LSP (Language Server Protocol) should handle
most of the heavy lifting. I've created my own plugin for my very basic
use case.
See: https://github.com/AdeAttwood/nrepl.nvim
The Ollama plugin provides a simple way to interact with an Ollama
language model from within Neovim. It uses as many of the built in
features as possible like buffers, tabs and markdown. This will make
this as small as possible so you don't have to install plugins with
loads of features you don't use.
Move the linters to run on BufReadPost instead of BufEnter. This was
running the linters whenever a buffer opened, when jumping around
quickly with <CTRL-i> and <CTRL-o> this would cause the linters to run a
lot when there were not needed. Now were are only running them once when
the buffer is read from disk.
Adds a custom function that will add cspell to all of the filetypes.
This will then only run `try_lint` once to improve the performance. Now
neoformat has gone fully async. When calling `try_lint` one after the
other, the first processes were getting canceled.
The jest out put is quite hard to process. This points that were add
into the quickfix list where often incorrect. This is a bit pointless
and I would find myself running it again in the terminal to see all of
the output.
This will have todo until I can find time to create proper test runner,
I really don't like the look of any of the plugins available.
Solargraph is best used installed in your Gemfile, this will allow you
to use manage the versions by each project. This will ensure that we are
using the version in your Gemfile, not the one installed globally.
Sets code folding of diffs and hunks so we can review hunks on there
own. This is good for larger pull requests, you can focus on the hunk
and not get overwhelmed by the rest of the code.
Adds a go to definition mapping for prr lines. This will allow you to go
straight to the line of code when reviewing a pull request. Seeing the
code in full context is really helpful when reviewing. You are then
there in your editor with all the linters and the LSP.
You can now run Jest on different parts of our app and populate the
quickfix list with the errors. You can run the tests on
- The current tests under the cursor
- All the tests in the current file
- All the tests on the project
This was an issue when editing files that don't have an lsp server. The
document highlight is only enabled for lsp's. This splits out the
highlighting and the diagnostic float into two separate auto commands.
The document highlight is now only enabled when the lsp is registered.
The diagnostic float is now enabled for all files. This is because the
the vim diagnostics are used for lots of other things like spell checking
and codelimate.
We have the kind of the completion item in the menu as an icon and the
name at the end. Now the end one has been replaced with the source name
so we can see what the completion item is and where it is from.
Due to starting to learn clojure, I've had java classes popping up in
the completion menu. Because they are so long it needs to be wider so I
can see then.
Install all of the tools and config for clojure development. This adds
the first configz module that is not yet properly integrated and
automaticaly run with the `dotfiles` shell command.
Now when loading vim the them from the shell is getting used rather than
only every using tomorrow night. When I use tomorrow it will be
inherited in vim.
Now when the editor is idle it will use the lsp to highlight other
references to the token under the cursor.
It will also show all the diagnostics in a popup like the lsp hover that
I manually use via <C-k>. This is helpful with long diagnostic messages
that run off the edge of the terminal when.
The biggest change here is that I am no longer cycling though completion
items as the first thing. Tab will now mainly control the snippet jumps
in completion not items. Completion items can be used exclusively with
<C-n> and <C-p> and <CR> to complete the selected item. If the
completion menu is open and <CR> is pressed, this will select and
complete the first item.
I will soon be able to remove the <C-h> and <C-l> mappings when my
muscle memory adapts as this will now be controlled with tab.
This moves all of the snippets into a yasnippet file format so I can
better manage and edit the snippets. I did not like having them all in
one file stored as strings.
This implements a custom file parser that will convert the snippet file
into a luasnip parsed snippet that uses the LSP snippet syntax.
It also ports over some of my most used snippets from the emacs config,
maybe one day I could share the snippet in vim and emacs.
This is from the dotfiles a long time ago, its still in vimscript, don't
think I will ever convert it over to lua.
This also disables base16 shell when the terminal is running inside vim
like I am doing with emacs.
I am finding my self wanting to view the output of command nowadays,
rather then wanting to run a quick command and see if its passed or not.
By removing the exit at the end of the command the terminal dose not
close now. If I want the terminal to close I can still append the
command with `&& exit` when using `Run`
This will now use a custom config file `$HOME/.cspell.json` that will
use a custom dictionary for my own words. This will have things like my
name so cspell stop telling me I spelt my name wrong.
Will also add the `--language-id` flag with the current filetype, so it
can match language specific words.
Currently it is set up to use tab to jump between placeholders and go to
the next completion item. This is causing conflicts when typing
variables in the snippet placeholders.
Now you can use <CTRL-h> to jump to the next placeholders. This is only
available when the completion is active so, dose not conflict with
switching windows.
