Diagnostic indicators are based only on messages from the "DiagnosticChanged" event, not from the entire buffer

[Yu-Leo]

Description

Diagnostic indicators are based only on messages from the "DiagnosticChanged" event, not from the entire buffer

I use nvim for Go (GoLang) development. I set up the lsp and linter using https://github.com/mfussenegger/nvim-lint.

I created my own autocmd: when saving the buffer (BufWritePost event) using the plugin https://github.com/mfussenegger/nvim-lint the linter is started, and its messages are added to diagnostics for the current buffer with the warning level.

In my case, it works like this:

1. I save the current buffer
2. LSP adds diagnostics
3. The linter is triggered
4. The linter adds diagnostics

When updating diagnostics, nvim-tree processes only those diagnoses that are contained in the "DiagnosticChanged" event. This is a problem for me because the "DiagnosticChanged" event occurs two times: in step 2 and in step 4.

If the diagnostics added by the linter will not contain diagnostic with the error level, and in the nvim-tree settings I will specify:

severity = {
  min = vim.diagnostic.severity.ERROR,
  max = vim.diagnostic.severity.ERROR,
}

then the diagnostics will not be displayed at all for the file in the nvim-tree.

I think it's more correct to analyze the entire diagnostics list for the buffer, which can be obtained using the vim.diagnostic.get function.

Neovim version

NVIM v0.10.2
Build type: RelWithDebInfo
LuaJIT 2.1.1731601260

Operating system and version

Linux 5.15.173-1-MANJARO

Windows variant

No response

nvim-tree version

master

Clean room replication

Not provided

Steps to reproduce

Not provided

Expected behavior

No response

Actual behavior

No response