feat: improve --list-csv and --list output

[bcouetil]

Closes #1809

Changes

• allowFailure: render exit codes as [42,137] instead of just true/false in both --list and --list-csv
• needs: empty when not specified (job follows stage ordering), [] when explicitly set to no dependencies — in both --list and --list-csv
• description column: always shown in --list, never in --list-csv (stable CSV structure)
• Shared private helpers formatAllowFailure and formatNeeds to avoid logic duplication

Context

These changes improve the usefulness of --list-csv for automated testing of GitLab CI job rules.

Note: the Windows id -u fix (originally in this PR) has been moved to a dedicated PR for #1684.

[cubic-dev-ai]

1 issue found across 8 files

Prompt for AI agents (unresolved issues)

Check if these issues are valid — if so, understand the root cause of each and fix them. If appropriate, use sub-agents to investigate and fix each issue separately.


<file name="src/commander.ts">

<violation number="1" location="src/commander.ts:213">
P2: `hasDescriptions` is derived from unfiltered jobs, so `--list` can incorrectly print a `description` column based only on hidden `when: never` jobs.</violation>
</file>

---

Since this is your first cubic review, here's how it works:

• cubic automatically reviews your code and comments on bugs and improvements
• Teach cubic by replying to its comments. cubic learns from your replies and gets better over time
• Add one-off context when rerunning by tagging @cubic-dev-ai with guidance or docs links (including llms.txt)
• Ask questions if you need clarification on any suggestion

_{Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.}

[firecow]

Having "unstable/non-deterministic" columns in CSV seems like a anti-pattern for people using CSV parsers doesn't it? @bcouetil

I can understand that --list doesn't need to show "description" column, its for humans after all.

hasDescriptions is computed before when: never filtering. If only a never job has a description, --list still shows the column for visible jobs that don't use it.

This id -u fix really should be in a PR of its own.

[bcouetil]

Hello, nice to talk to you once again 😊

Having "unstable/non-deterministic" columns in CSV seems like a anti-pattern for people using CSV parsers doesn't it? @bcouetil
I can understand that --list doesn't need to show "description" column, its for humans after all.

You meant --list-csv I guess. I agree, I will remove this column entirely in this case.

hasDescriptions is computed before when: never filtering. If only a never job has a description, --list still shows the column for visible jobs that don't use it.

I wanted to have the column showing independently from activated jobs. In fact, consistency for PEOPLE using the descriptions, and consistency for those not using it. I guessed a on/off usage on a project... I personally do not use it, at all. So my guess is far from perfect. Please tell me your final wish on this.

This id -u fix really should be in a PR of its own.

OK, PR incoming 🫡

[firecow]

Looks good — description is already dropped from --list-csv in this diff, which addresses my concern. The hasDescriptions/--list point is out of scope for this PR then.

[bcouetil]

Looks good — description is already dropped from --list-csv in this diff, which addresses my concern. The hasDescriptions/--list point is out of scope for this PR then.

Sorry, I saw your comment while adding the same to --list... I should not have squashed my commits, but I wanted to share the logic between the too.

I kept description on --list, but added both allow_failure and needs change, to have maximum info.

What do you think @firecow ?

[bcouetil]

I've now rebased. I think it is finished on my side.