[idea] Wire vim.lsp.formatexpr() so gq uses the LSP/conform formatter #127

What

Set formatexpr in LspAttach to v:lua.vim.lsp.formatexpr() so that the gq operator delegates to the attached LSP (or conform, which the config already wires as LSP fallback) instead of the built-in line-breaking algorithm.

Where

lua/config/plugin_config.lua — LspAttach autocmd, lines 104–115.

Why it matters

The config already invests in formatting quality: conform.nvim with stylua, ruff_format, prettier; default_format_opts = { lsp_format = "fallback" }; and a custom <leader>lf keymap. But gq (reformat operator / motion) still falls back to Vim's dumb line-wrapper, ignoring all of that. For prose-heavy work (Markdown notes via obsidian.nvim, inline comments) gqq / gqip / gq<motion> are natural muscle-memory, and they currently produce inconsistent results vs <leader>lf.

Neovim 0.11+ ships vim.lsp.formatexpr() exactly for this purpose.

Suggested implementation

Inside the LspAttach callback (alongside the existing gd keymap):

vim.api.nvim_create_autocmd("LspAttach", {
    callback = function(ev)
        local client = vim.lsp.get_client_by_id(ev.data.client_id)
        -- existing completion + gd setup …
        vim.bo[ev.buf].formatexpr = "v:lua.vim.lsp.formatexpr()"
    end,
})

vim.lsp.formatexpr() honours the LSP textDocument/rangeFormatting method when the attached server supports it; otherwise it falls back to the built-in algorithm — so there is no regression for buffers with no LSP attached.

Note: this interacts with formatoptions. If the t or c flags are set (they are currently absent, which is correct), the formatexpr fires on auto-wrap too. With the current rqnl1j flags, formatexpr only fires on explicit gq calls, which is the desired behavior.