add ws diagnose util to detect cyclic dependencies and multiple variants of the same dependency by iurimatias · Pull Request #17 · logos-co/logos-workspace

iurimatias · 2026-03-20T15:39:20Z

No description provided.

…nts of the same dependency

Copilot

Pull request overview

This PR adds a new ws nix-diagnose command to help troubleshoot Nix flake dependency problems (cycles and duplicate versions) and documents the new workflow in the repo docs.

Changes:

Add ws nix-diagnose <repo> command to analyze flake.lock for circular deps and duplicate dependency revisions.
Register nix-diagnose in ws help/dispatch and expand documentation in README.md and CLAUDE.md.
Add logos-nix to the workspace repo registry.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 8 comments.

File	Description
`scripts/ws`	Adds `cmd_nix_diagnose`, registers the new command, and extends the REPOS registry.
`README.md`	Documents the new `ws nix-diagnose` command and its intended output/fixes.
`CLAUDE.md`	Updates the command quick-reference with the new diagnostics command.

Copilot · 2026-03-20T15:44:49Z

scripts/ws

+        select(.value | type == "string") |
+        .value as $child |


The jq extraction treats dependency edges as only those inputs whose value is a string, but in this repo’s flake.lock many inputs are arrays (e.g. ["nixpkgs"] or longer paths). This will drop most edges and can incorrectly report “No circular dependencies found”. Update the jq to also handle array inputs (e.g. map arrays to their first element / target node key) so the dependency graph is complete.

Suggested change

select(.value | type == "string") |

.value as $child |

.value as $input |

(if ($input | type) == "string" then

$input

elif ($input | type) == "array" and ($input | length) > 0 then

$input[0]

else

empty

end) as $child |

Copilot · 2026-03-20T15:44:50Z

scripts/ws

+        .nodes.root.inputs | to_entries[] |
+        select(.value | type == "string") | .key
+      ' "$lock_file" 2>/dev/null | head -1 || true)
+      # Visit all repos reachable from root inputs
+      local root_inputs
+      root_inputs=$(jq -r '
+        .nodes.root.inputs | to_entries[] |
+        select(.value | type == "string") | .value


Root inputs are filtered to type == "string", but root inputs can also be arrays in flake.lock (path-form). That can make root_inputs incomplete/empty and skip traversal entirely. Consider normalizing input refs so both string and array forms are traversed.

Suggested change

.nodes.root.inputs | to_entries[] |

select(.value | type == "string") | .key

' "$lock_file" 2>/dev/null | head -1 || true)

# Visit all repos reachable from root inputs

local root_inputs

root_inputs=$(jq -r '

.nodes.root.inputs | to_entries[] |

select(.value | type == "string") | .value

.nodes.root.inputs

| to_entries[]

| .value

| if type == "string" then .

elif type == "array" and length > 0 then .[0]

else empty

end

' "$lock_file" 2>/dev/null | head -1 || true)

# Visit all repos reachable from root inputs

local root_inputs

root_inputs=$(jq -r '

.nodes.root.inputs

| to_entries[]

| .value

| if type == "string" then .

elif type == "array" and length > 0 then .[0]

else empty

end

Copilot · 2026-03-20T15:44:50Z

scripts/ws

+    select(.value | type == "string") |
+    $parent + " " + .value + " " + .key


The adjacency list for _find_path only includes inputs whose value is a string. In flake.lock, many inputs are arrays, so paths to conflicting nodes will often show as “(unreachable)” even when reachable. Normalize .value so both string and array forms contribute edges.

Suggested change

select(.value | type == "string") |

$parent + " " + .value + " " + .key

.key as $input_name |

.value as $v |

(if ($v | type) == "string" then [$v]

elif ($v | type) == "array" then $v

else [] end)[] as $child |

$parent + " " + $child + " " + $input_name

Copilot · 2026-03-20T15:44:50Z

scripts/ws

+  local input_name dir_name
+  input_name=$(get_field "$entry" 1)


input_name is computed but never used in this command. Remove it (and the assignment) or use it in output so the function doesn’t carry dead variables.

Suggested change

local input_name dir_name

input_name=$(get_field "$entry" 1)

local dir_name

Copilot · 2026-03-20T15:44:50Z

scripts/ws

+      local start_repo
+      start_repo=$(jq -r '
+        .nodes.root.inputs | to_entries[] |
+        select(.value | type == "string") | .key
+      ' "$lock_file" 2>/dev/null | head -1 || true)


start_repo is computed but never used. This looks like leftover scaffolding; consider removing it to avoid confusion and keep the cycle detection logic focused.

Suggested change

local start_repo

start_repo=$(jq -r '

.nodes.root.inputs | to_entries[] |

select(.value | type == "string") | .key

' "$lock_file" 2>/dev/null | head -1 || true)

Copilot · 2026-03-20T15:44:50Z

scripts/ws

+        local deps_str
+        deps_str=$(echo "$graph_data" | grep "^${node}|" | head -1 | cut -d'|' -f2 || true)
+        local dep
+        for dep in ${deps_str//,/ }; do
+          [[ -z "$dep" ]] && continue
+          local state
+          state=$(grep "^${dep} " "$_dfs_state_file" 2>/dev/null | tail -1 | cut -d' ' -f2 || true)
+          if [[ "$state" == "visiting" ]]; then


These grep patterns interpolate repo/node names directly into a regex (e.g. ^${node}|). Repo IDs / node keys can contain characters like . that are meaningful in regex, which can cause incorrect matches. Prefer fixed-string matching (grep -F) or escape the variables before using them in regex patterns.

Copilot · 2026-03-20T15:44:51Z

scripts/ws

+    repo_edges=$(jq -r '
+      .nodes as $nodes |
+      [ $nodes | to_entries[] |
+        select(.value.inputs != null) |
+        .key as $parent |
+        ( $nodes[$parent].original.repo // ($parent | sub("_[0-9]+$"; "")) ) as $src |
+        .value.inputs | to_entries[] |
+        select(.value | type == "string") |
+        .value as $child |
+        ( $nodes[$child].original.repo // ($child | sub("_[0-9]+$"; "")) ) as $dst |
+        select($src != $dst) |
+        $src + " " + $dst
+      ] | unique[]
+    ' "$lock_file" 2>/dev/null || true)
+
+    if [[ -z "$repo_edges" ]]; then
+      echo -e "  ${GREEN}No circular dependencies found${NC}"
+    else


In the cycle section, jq errors are suppressed (2>/dev/null || true), which can turn parse failures into an empty edge list and a misleading “No circular dependencies found”. Consider checking jq’s exit status and reporting a clear parse/error message instead of silently continuing.

Copilot · 2026-03-20T15:44:51Z

scripts/ws

 # has_flake: "yes" or "no"

 REPOS=(
+  # Shared Nix infrastructure


logos-nix is added to the REPOS registry, but it isn’t present in .gitmodules yet. With this change, ws init will try to git submodule add it in user workspaces, which creates local .gitmodules/index changes. Consider adding the submodule entry to .gitmodules in this PR (or documenting why it should remain auto-added).

Suggested change

# Shared Nix infrastructure

# Shared Nix infrastructure

# NOTE: logos-nix is intentionally auto-added as a submodule by `ws init`

# and is not predeclared in .gitmodules; this keeps it workspace-local.

add ws diagnose util to detect cyclic dependencies and multiple varia…

9cd747a

…nts of the same dependency

Copilot AI review requested due to automatic review settings March 20, 2026 15:39

iurimatias merged commit 90a75a4 into master Mar 20, 2026
2 of 3 checks passed

Copilot started reviewing on behalf of iurimatias March 20, 2026 15:40 View session

Copilot AI reviewed Mar 20, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

add ws diagnose util to detect cyclic dependencies and multiple variants of the same dependency#17

add ws diagnose util to detect cyclic dependencies and multiple variants of the same dependency#17
iurimatias merged 1 commit intomasterfrom
docs/nix-diagnose

iurimatias commented Mar 20, 2026

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 20, 2026

Uh oh!

Copilot AI Mar 20, 2026

Uh oh!

Copilot AI Mar 20, 2026

Uh oh!

Copilot AI Mar 20, 2026

Uh oh!

Copilot AI Mar 20, 2026

Uh oh!

Copilot AI Mar 20, 2026

Uh oh!

Copilot AI Mar 20, 2026

Uh oh!

Copilot AI Mar 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

-        select(.value | type == "string") |
-        .value as $child |
+        .value as $input |
+        (if ($input | type) == "string" then
+           $input
+         elif ($input | type) == "array" and ($input | length) > 0 then
+           $input[0]
+         else
+           empty
+         end) as $child |

		select(.value \| type == "string") \|
		$parent + " " + .value + " " + .key

-    select(.value | type == "string") |
-    $parent + " " + .value + " " + .key
+    .key as $input_name |
+    .value as $v |
+    (if ($v | type) == "string" then [$v]
+     elif ($v | type) == "array" then $v
+     else [] end)[] as $child |
+    $parent + " " + $child + " " + $input_name

	local input_name dir_name
	input_name=$(get_field "$entry" 1)
	local dir_name

Conversation

iurimatias commented Mar 20, 2026

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 20, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants