Make formatexpr use the same option as conform.format()

[xulongwu4]

The options for formatexpr is different than those used in conform.format(). This discrepancy causes issues when one sets up filetype-specific values for lsp_format or timeout_ms. For example, I use the following values for formatters_by_ft:

formatters_by_ft = {["_"] = { "trim_newlines", "trim_whitespace", lsp_format = "prefer" }}

I also format buffers on save.

Now when I edit a .C file and run wq, clangd kicks in as I set lsp_format = "prefer". The file is formatted by clangd and saved correctly. Everything is good.

Now if I try to use gq, the format is performed by trim_newlines and trim_whitespace, because the default values for lsp_format in formatexpr is fallback. This inconsistency causes confusion when using conform. See also #824 #408 #752

This PR resolves this issue so that formatexpr will respect options used by formatters_by_ft.

[Copilot]

Pull request overview

This PR makes formatexpr use the same option defaults as conform.format() by removing the hardcoded timeout_ms = 500 and lsp_format = "fallback" values. This allows formatexpr to respect filetype-specific options configured in formatters_by_ft and default_format_opts, resolving inconsistencies where gq formatting would use different settings than format-on-save.

Key changes:

• Removed hardcoded timeout_ms and lsp_format defaults from formatexpr function
• formatexpr now inherits options from formatters_by_ft configuration, default_format_opts, and finally the hardcoded defaults in conform.format()

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[stevearc]

Hmmm...yes I think that behavior predates the default_format_opts and some of the other methods of configuring formatting options by filetype. Doesn't really make sense to override those values anymore. Thanks for the PR!