Problem:
:checkhealth shows a confusing message if user sets
g:loaded_python3_provider=1.
Solution:
- Show a warning if that var is set to 1.
- Update provider modules to default to 0. Any user code that is
checking for 1, is like already broken because these may be set to 2.
(cherry picked from commit b4906577c9)
Co-authored-by: Sathya Pramodh <94102031+sathya-pramodh@users.noreply.github.com>
The benefit of this is that users only pay for what they use. If e.g.
only `vim.lsp.buf_get_clients()` is called then they don't need to load
all modules under `vim.lsp` which could lead to significant startuptime
saving.
Also `vim.lsp.module` is a bit nicer to user compared to
`require("vim.lsp.module")`.
This isn't used for some nested modules such as `filetype` as it breaks
tests with error messages such as "attempt to index field 'detect'".
It's not entirely certain the reason for this, but it is likely it is
due to filetype being precompiled which would imply deferred loading
isn't needed for performance reasons.
Problem: ruby#Detect() and node#Detect() don't return a [prog, err] pair
which means callers must special-case them.
Solution: align their return signatures with the perl/pythonx providers.
Instead of deciding provider status in eval_has_provider, move the
decision to the provider Vim scripts.
Previously, provider loading worked as follows:
1. eval_has_provider() verified provider availability by searching for
the provider#providername#Call function and cached this verificaion as a static
variable for some providers
2. providers short-circuited on loading to prevent the definition of the
Call function (with the exception of the node provider that did not)
This commit changes the expected interface between nvim and its
providers to facilitate provider reloading, by splitting the
verification of the provider from the availability of the Call function.
eval_has_provider() now checks for a provider#providername#enabled
variable. It is up to the provider script to set this to 0 or 1
accordingly. eval_call_provider() remains unchanged.
All providers hosting a Call function were updated to respect this.
The clipboard provider now has a Reload function to reload the
provider.
Before this commit, if user does this:
let g:node_host_prog = '~/.nvm/versions/node/v11.3.0/bin/neovim-node-host'
the "~/" is not expanded to user's home directory.
`:help g:ruby_host_prog` suggests a path with "~/" so technically we
already claimed to support this.
closes https://github.com/neovim/node-client/issues/102
When rbenv is used for managing Ruby installations, just checking for an
executable called "neovim-ruby-host" is not enough. It has to be run as well.
If it does not return 0, then neovim-ruby-host is merely a shim for another Ruby
installation.
$ rbenv versions
2.5.0
* 2.5.1 (set by /Users/mhi/.rbenv/version)
$ rbenv whence neovim-ruby-host
2.5.0
$ which neovim-ruby-host
/Users/mhi/.rbenv/shims/neovim-ruby-host
$ neovim-ruby-host
rbenv: neovim-ruby-host: command not found
The `neovim-ruby-host' command exists in these Ruby versions:
2.5.0
$ echo $?
127
Additionally, the detection logic was moved from provider#ruby#Detect() to
s:detect(), because the former is run in the sandbox which forbids calling
system().
ruby uses batchfiles with 'cmd' extension.
gem creates batchfiles with 'bat' extension.
`gem install rails` does the following in Windows (not Cygwin):
1. Run `gem.cmd install rails` on cmd.exe
2. gem.cmd runs `ruby.exe -x gem install rails`
3. `rails` gem is installed.
`rails.bat` is created in the same directory
where ruby.exe and gem.cmd reside.
neovim-ruby-host is a ruby script.
neovim-node-host is a shell script.
Both don't work in cmd.exe so gem and npm provide batchfile shims.
Return the full path of these shims, cmd.exe knows better what to do with these files.
This allows users who have per-project Ruby versions (e.g. with `rvm`)
to pin to a particular gem installation.
For example: `let g:ruby_host_prog = 'rvm system do neovim-ruby-host'`
The exists('g:loaded_foo') convention is rather common, and may be
relied upon in some cases. It's also very unlikely that a user or plugin
has any reason to set g:loaded_foo to zero, so the principle of least
surprise can be brushed aside here.
https://github.com/neovim/neovim/issues/6107#issuecomment-279532143
