pkgs/profpatsch/nman: ✨ add --file flag to use local nix file by Profpatsch · Pull Request #59 · openlab-aux/vuizvui

Profpatsch · 2025-07-03T14:36:48Z

Add support for using ./default.nix instead of when the --local or -l flag is specified. This allows viewing man pages for packages defined in local development environments or custom package sets.

Changes:

Add use_local field to Main struct
Parse --local/-l command line option
Conditionally use ./default.nix in Nix expression generation
Update help text and man page documentation with new option and examples
Maintain backward compatibility (default behavior unchanged)

🤖 Generated with Claude Code

Add support for using ./default.nix instead of <nixpkgs> when the --local or -l flag is specified. This allows viewing man pages for packages defined in local development environments or custom package sets. Changes: - Add use_local field to Main struct - Parse --local/-l command line option - Conditionally use ./default.nix in Nix expression generation - Update help text and man page documentation with new option and examples - Maintain backward compatibility (default behavior unchanged) 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>

When using dotted attribute paths like `pkgs.profpatsch.nman`, nman was incorrectly trying to find a man page named "pkgs.profpatsch.nman" instead of extracting the actual page name "nman" from the last component. Changes: - Add extract_page_name_from_attr() helper that splits on '.' and returns last component - Update CLI parsing to use helper for auto-detected page names (cases 2 and 3 args) - Add comprehensive test coverage for the helper function - Maintain backward compatibility for simple attribute names and explicit page names - Add description to the man page Now `nman --local pkgs.profpatsch.nman` correctly builds the full attribute but searches for the "nman" man page instead of "pkgs.profpatsch.nman". 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>

sternenseemann

I think the specific --local flag is a bit silly. I'd prefer just adding -f/--file like the new Nix/Lix CLI does. Then you'd write nman -f . mypkg.

Profpatsch · 2025-08-04T15:27:31Z

I special-cased it for . now, since I have no clue about the nix -f semantics and really kinda don’t want to know atm haha

Refactor the inflexible --local flag to use -f/--file <path> syntax, matching modern Nix CLI conventions. This allows specifying any .nix file instead of hardcoding ./default.nix. Usage: nman -f . mypkg (equivalent to old --local behavior) nman -f /path/to/file.nix mypkg (new flexibility) 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>

Unify the logic for determining the import path and generate the Nix expression only once, reducing code duplication and making the logic clearer. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>

sternenseemann · 2025-08-10T11:44:45Z

 .Nm
 into verbose mode, where it prints all commands it executes.
+.It Fl -file Ar path | f Ar path
+Use the specified .nix file


Suggested change

Use the specified .nix file

Use the Nix expression at

No it only allows nix files? I think. We pass it to import but if we want to allow arbitrary expressions we’d have to escape them properly first

It is clear that it's a file from what follows.

sternenseemann · 2025-08-10T11:57:02Z

+    let mut positional_args = Vec::new();
+
+    let mut i = 1; // Skip program name
+    while i < all_args.len() {


This is a bit icky.

Not sure how to do it better tbh

Does while let Some(arg) = … entitle us to call next() inside the loop? Surely something like that must be possible.

sternenseemann · 2025-08-10T11:58:35Z

+    }
+
+    // Only process positional arguments if we haven't already set an error
+    if let ShowUsage { err_msg: Some("Unexpected number of arguments") } = cli_res {


We are matching an error message to detect that no error occurred? This is extremely brittle and also super confusing code.

Yeah it should not match on the exact error, any error should skip

Newer rust versions apparently can specify the extern directly without the link annotation (was a warning).

This should pass the right kinds of path to nix in most situations.

Replace tuple-style CliAction::Man with inline struct fields for better readability and maintainability. Move file_path from Main struct to the action itself for better separation of concerns. Changes: - Use named fields (attr, section, page, file_path) instead of positional args - Keep efficient &'a str lifetimes for all string references - Pass file_path explicitly to open_man_page method - Remove file_path from Main struct 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <noreply@anthropic.com>

When we match positional args, we want to skip whenever we are already in the ShowUsage case.

sternenseemann · 2025-08-21T19:07:30Z


-#[link(name = "c")]
-extern {
+extern "C" {


Wouldn't it be better to be explicit?

Suggested change

extern "C" {

#[link(name = "c")]

extern "C" {

sternenseemann · 2025-08-21T19:13:55Z

-        //               like <vuizvui>, /home/lukas/src/nix/nixpkgs, …
+        let nix_import = match file_path {
+            Some(path) => {
+                if path.starts_with('/') {


Should also check if the path matches ~/… or <…> which are also valid Nix paths.

I guess we should also clarify in the man page that we allow this special path syntax.

Could be nice to add a check whether it is unnecessary to add the ./ prefix, but it also doesn't really matter.

sternenseemann · 2025-08-21T19:33:53Z

        attr: &'a str,
        section: Option<&'a str>,
        page: &'a str,
+        file_path: Option<&'a str>,


Maybe we should clarify that we treat this as a subset of Nix syntax which is passed on, hence we don't use Path.

sternenseemann · 2025-08-21T19:45:03Z

+    }
+
+    // Only process positional arguments if we haven't already set an error
+    if let ShowUsage { err_msg: _ } = cli_res {


but this matches if we have set an error already!!! you've made it worse!!!!

You could also return a Result from either main or a dedicated function and return early instead of adding your own Resultish type. For example if you want to have different errors with different actions, you could still use Result<_, CustomErrorEnum>.

sternenseemann · 2025-08-21T19:46:27Z

+
+    // Only process positional arguments if we haven't already set an error
+    if let ShowUsage { err_msg: _ } = cli_res {
+        cli_res = match positional_args.len() {


Can't you match on a slice of positonal_args?

Profpatsch and others added 2 commits July 3, 2025 16:36

Profpatsch force-pushed the nman-improvements branch from a192a6f to ffb8977 Compare July 3, 2025 17:14

sternenseemann reviewed Aug 2, 2025

View reviewed changes

Comment thread pkgs/profpatsch/nman/nman.rs Outdated

Comment thread pkgs/profpatsch/nman/nman.rs

Profpatsch and others added 2 commits August 4, 2025 17:31

Profpatsch force-pushed the nman-improvements branch from 160d708 to 76c7c94 Compare August 4, 2025 15:31

Profpatsch changed the title ~~pkgs/profpatsch/nman: ✨ add --local flag to use local default.nix~~ pkgs/profpatsch/nman: ✨ add --file flag to use local nix file Aug 4, 2025

sternenseemann requested changes Aug 10, 2025

View reviewed changes

Profpatsch added 2 commits August 15, 2025 14:05

pkgs/sternenseemann/temp: use extern "C"

3b1820e

Newer rust versions apparently can specify the extern directly without the link annotation (was a warning).

pkgs/Profpatsch/nman: allow any absolute or relative paths for -f

0aafc7f

This should pass the right kinds of path to nix in most situations.

Profpatsch force-pushed the nman-improvements branch from bcd41b6 to 0aafc7f Compare August 15, 2025 12:11

Profpatsch and others added 2 commits August 15, 2025 14:37

pkgs/profpatsch/nman: Don’t match on specific usage

f7c85c5

When we match positional args, we want to skip whenever we are already in the ShowUsage case.

Profpatsch force-pushed the nman-improvements branch from 4539d97 to f7c85c5 Compare August 15, 2025 12:38

sternenseemann reviewed Aug 21, 2025

View reviewed changes

Conversation

Profpatsch commented Jul 3, 2025

Uh oh!

sternenseemann left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Profpatsch commented Aug 4, 2025

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants