feat(diagnostic): add `vim.diagnostic.count()`
Problem: Getting diagnostic count based on the output of
`vim.diagnostic.get()` might become costly as number of diagnostic
entries grows. This is because it returns a copy of diagnostic cache
entries (so as to not allow users to change them in place).
Getting information about diagnostic count is frequently used in
statusline, so it is important to be as fast as reasonbly possible.
Solution: Add `vim.diagnostic.count()` which computes severity
counts without making copies.
As specified by MAINTAIN.md, features should be soft deprecated at first
(meaning no warnings) to give people a chance to adjust. The problem
with this approach is that deprecating a feature becomes harder than
usual as during the soft deprecation period you need to remember not to
issue a warning, and during the hard deprecation period you need to
remember to start issuing a warning.
This behavior is only enforced if the `plugin` parameter is `nil` as
plugins may not want this specific behavior.
refactor(lsp): move glob parsing to vim.glob
Moving the logic for using vim.lpeg to create a match pattern from a
glob into `vim.glob`. There are several places in the LSP spec that
use globs, and it's very useful to have glob matching as a
generally-available utility.
Diagnostic signs should now be configured with vim.diagnostic.config(),
but "legacy" sign definitions should go through the standard deprecation
process to minimize the impact from breaking changes.
This is the command invoked repeatedly to make the changes:
:%s/^\(.*\)|\%(\*\(\d\+\)\)\?$\n\1|\%(\*\(\d\+\)\)\?$/\=submatch(1)..'|*'..(max([str2nr(submatch(2)),1])+max([str2nr(submatch(3)),1]))/g
Problem: Many places in the code use `findoption()` to access an option using its name, even if the option index is available. This is very slow because it requires looping through the options array over and over.
Solution: Use option index instead of name wherever possible. Also introduce an `OptIndex` enum which contains the index for every option as enum constants, this eliminates the need to pass static option names as strings.
When we convert a Lua table to an Object, we consider the table a
"dictionary" if it contains only string keys, and an array if it
contains all numeric indices with no gaps. While rare, Lua tables can
have both strictly numeric indices and gaps (e.g. { [2] = 2 }). These
currently cannot be serialized because it is not considered an array.
However, we know the maximum index of the table and as long as all of
the keys in the table are numeric, it is still possible to serialize
this table as an array. The missing indices will have nil values.
Problem:
Empty string is a valid JSON key, but json_decode() treats an object
with empty key as ":help msgpack-special-dict". #20757
:echo json_decode('{"": "1"}')
{'_TYPE': [], '_VAL': [['', '1']]}
Note: vim returns `{'': '1'}`.
Solution:
Allow empty string as an object key.
Note that we still (currently) disallow empty keys in object_to_vim() (since 7c01d5ff92):
f64e4b43e1/src/nvim/api/private/converter.c (L333-L334)Fix#20757
Co-authored-by: Justin M. Keyes <justinkz@gmail.com>
Problem:
With vim.treesitter.foldexpr, `o`-ing two lines above a folded region
opens the fold. This does not happen with legacy foldexprs. For example,
make a markdown file with the following text (without indentation),
enable treesitter fold, and follow the instruction in the text.
put cursor on this line and type zoo<Esc>
initially folded, revealed by zo
# then this fold will be opened
initially folded, revealed by o<Esc>
Analysis:
* `o` updates folds first (done in `changed_lines`), evaluating
foldexpr, and then invokes `on_bytes` (done in `extmark_splice`).
* Treesitter fold allocates the foldinfo for added lines (`add_range`)
on `on_bytes`.
* Therefore, when treesitter foldexpr is invoked while running `o`, it
sees outdated foldinfo.
Solution:
`extmark_splice`, and then `changed_lines`. This seems to be the
standard order in other places, e.g., `nvim_buf_set_lines`.
Quick fix as follow up to https://github.com/neovim/neovim/pull/26108
kqueue only reports events on a watched folder itself, not for files
created or deleted within. So the approach the PR took doesn't work on FreeBSD.
We'll either need to bring back polling for it, combine watching with manual
file tracking, or disable LSP file watching on FreeBSD
Should help with https://github.com/neovim/neovim/issues/23291
On linux `new_fs_event` doesn't support recursive watching, but we can
still use it to watch folders.
The downside of this approach is that we may end up sending some false
`Deleted` events. For example, if you save a file named `foo` there will
be a intermediate `foo~` due to the save mechanism of neovim.
The events we get from vim.uv in that case are:
- rename: foo~
- rename: foo~
- rename: foo
- rename: foo
- change: foo
- change: foo
The mechanism in this PR uses a debounce to reduce this to:
- deleted: foo~
- changed: foo
`foo~` will be the false positive.
I suspect that for the LSP case this is good enough. If not, we may need
to follow up on this and keep a table in memory that tracks available
files.
The prefix option of the diagnostic virtual text can be a function,
but previously it was only a function of diagnostic.
This function should also have additional parameters index and total,
more consistently and similarily as in the prefix function for
`vim.diagnostic.open_float()`.
These additional parameters will be useful when there are too many
number of diagnostics in a single line.
Problem: Peeking and flushing output slows down execution.
Solution: Do not update the mode message when global_busy is set. Do not
flush when only peeking for a character. (Ken Takata)
cb574f4154
When tabstop and shiftwidth are not equal, tabs are inserted as individual
spaces and then rewritten as tab characters in a second pass. That second pass
did not call changed_bytes which resulted in events being omitted.
Fixes#25092
