fix(IW-556): keep the developer's tmux config out of the plumbing test #574

Merged

mph merged 17 commits from IW-556 into main 2026-09-16 15:21:24 +00:00

Conversation 0 Commits 17 Files changed 6 +691 -1

claude commented 2026-09-16 13:36:57 +00:00

Member

Copy link

Summary

test/start.bats case "the tmux session does not carry iw's internal plumbing" failed on a developer machine at every run: IW_HOME was absent from the pane.

Cause. -L <socket> isolates the tmux socket, not the config. The test's server still read ~/.config/tmux/tmux.conf, and that config runs set-environment -gu IW_HOME on purpose, so fish supplies a live value. iw was correct; the test depended on the developer's config.

Fix. The case runs iw start with HOME="$TEST_PARENT", which holds no tmux or fish config. TmuxAdapter does not change.

Rejected:

• new-session -e IW_HOME=… in TmuxAdapter: it freezes IW_HOME in each session, which that config prevents on purpose.
• XDG_CONFIG_HOME: tmux 3.6a still reads ~/.config/tmux/tmux.conf.
• A test-started server with -f /dev/null: the pane would hold the test's environment, so the case would no longer test sessionEnv.

Defects

• Fixed: the false red in start.bats (Defect 1 of the analysis).
• Not fixed, filed as #572 (low priority): iw start on a tmux server that already runs gives the pane that server's IW_* values (Defect 2).
• Seen again and noted on #200: review-state update --append-artifact duplicates an artifact.

Testing

• Before: the case failed at line 145. After: it passes.
• Mutations of sessionEnv at core/adapters/Tmux.scala:74: dropping IW_HOME fails the case (line 145); keeping every IW_* name fails it (line 142). Both reverted.
• bats test/start.bats: 6/6 pass. ./iw ./test e2e: exit 0. Pre-commit and pre-push hooks (format, scalafix, compile with warnings as errors, unit tests): pass.

Release notes are not included: the change touches a test only.

Closes #556

🤖 Generated with Claude Code

## Summary `test/start.bats` case "the tmux session does not carry iw's internal plumbing" failed on a developer machine at every run: `IW_HOME` was absent from the pane. **Cause.** `-L <socket>` isolates the tmux socket, not the config. The test's server still read `~/.config/tmux/tmux.conf`, and that config runs `set-environment -gu IW_HOME` on purpose, so fish supplies a live value. `iw` was correct; the test depended on the developer's config. **Fix.** The case runs `iw start` with `HOME="$TEST_PARENT"`, which holds no tmux or fish config. `TmuxAdapter` does not change. Rejected: - `new-session -e IW_HOME=…` in `TmuxAdapter`: it freezes `IW_HOME` in each session, which that config prevents on purpose. - `XDG_CONFIG_HOME`: tmux 3.6a still reads `~/.config/tmux/tmux.conf`. - A test-started server with `-f /dev/null`: the pane would hold the test's environment, so the case would no longer test `sessionEnv`. ## Defects - Fixed: the false red in `start.bats` (Defect 1 of the analysis). - Not fixed, filed as #572 (low priority): `iw start` on a tmux server that already runs gives the pane that server's `IW_*` values (Defect 2). - Seen again and noted on #200: `review-state update --append-artifact` duplicates an artifact. ## Testing - Before: the case failed at line 145. After: it passes. - Mutations of `sessionEnv` at `core/adapters/Tmux.scala:74`: dropping `IW_HOME` fails the case (line 145); keeping every `IW_*` name fails it (line 142). Both reverted. - `bats test/start.bats`: 6/6 pass. `./iw ./test e2e`: exit 0. Pre-commit and pre-push hooks (format, scalafix, compile with warnings as errors, unit tests): pass. Release notes are not included: the change touches a test only. Closes #556 🤖 Generated with [Claude Code](https://claude.com/claude-code)

claude added 16 commits 2026-09-16 13:36:57 +00:00

workflow(dx): Create diagnostic analysis for IW-556 ... 25af3436e0

start.bats "the tmux session does not carry iw's internal plumbing" fails at
baseline: IW_HOME missing from the captured pane env

Artifacts:
- analysis.md

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011M2HjEYoJpcfD43KYP6zxt

workflow(dx): Create task index for IW-556 ... 53664dcaa0

start.bats "the tmux session does not carry iw's internal plumbing" fails at
baseline: IW_HOME missing from the captured pane env

Artifacts:
- tasks.md
- review-state.json (tasks_ready)

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011M2HjEYoJpcfD43KYP6zxt

docs(IW-556): intake verdict — needs-human ... b55261c607

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_011M2HjEYoJpcfD43KYP6zxt

docs(IW-556): limit new-session -e to IW_* names (Decision D1) ... b7d1a9d2a1

The intake review showed that sending all of sessionEnv through
`new-session -e` freezes the secrets that tmux.conf removes on purpose.
Michal chose the narrow scope: IW_HOME, IW_TMUX_SOCKET and removal
entries for held-back IW_* names only.

Also add a BATS case that starts the server with -f, so the repository
holds Defect 1, record the cases the narrow scope leaves open (R5), and
correct the review's non-blocking notes.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

docs(IW-556): always remove the six plumbing names and IW_HOME ... 2237e835ec

Michal chose to close Risk R5 in this fix: the session gets a removal
entry for each of the six plumbing names and for IW_HOME even when the
iw process does not hold them, because a tmux server that iw did not
start can hold them.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

chore(IW-556): update review-state to tasks_ready 5d4f22dc2d

workflow(dx): Create task index for IW-556 ... 5cb4560cd2

start.bats "the tmux session does not carry iw's internal plumbing" fails at baseline: IW_HOME missing from the captured pane env

Task index regenerated after Decision D1 bounded the -e scope.

Artifacts:
- tasks.md
- review-state.json (tasks_ready)

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

chore(IW-556): update review-state 342c3d6f43

docs(IW-556): record review-state duplicate artifact follow-up ... 5ed2dfd456

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

docs(IW-556): empty-value removal (D2) and phases split by fix ... 3df423c8ed

Only `-e NAME=` changes the first pane, so Michal chose it (Decision D2).
One change fixes both defects, so Phase 1 now holds the whole fix with
all its tests first, and Phase 2 holds the tmux contract case and the
documentation. tasks.md also states the IW_HOME rule as D1 does.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Phase 1: context ready ... ba57bd7324

Issue: IW-556
Phase: 1 - `new-session -e` gives each session its `IW_*` state

This commit marks the checkpoint for phase 1 context generation.
Use dx-fix IW-556 to return to this point.

It also applies Michal's decision from the context review:
IW_TMUX_SOCKET follows the same set-or-remove rule as IW_HOME.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

chore(IW-556): update review-state to context_ready 5a7a8d5af1

fix(IW-556): keep the developer's tmux config out of the plumbing test ... 442feba79f

The case read a pane on a tmux server that loads ~/.config/tmux/tmux.conf,
because -L isolates the socket and not the config. A config that unsets
IW_HOME made the case fail on that machine only. iw start now runs with a
HOME that holds no config, so the pane shows what iw gave the session.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

docs(IW-556): reduce scope to the test ... 59f58449eb

Michal reduced IW-556 to the test: new-session -e would freeze IW_HOME,
which the developer's tmux config prevents on purpose. D1 and D2 are
marked superseded, Defect 2 moves to FOLLOW-UPS.md with low priority,
and the phase context and task index describe the one-line test fix.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

chore(IW-556): update review-state to implementing 6f7bfc6664

docs(IW-556): follow-ups filed (#572, comment on #200), phase 1 done ... 73ac5cd5d9

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

mph added 1 commit 2026-09-16 13:38:08 +00:00

chore(IW-556): update review-state to all_complete 4136b168f8

mph merged commit 9b90522d66 into main 2026-09-16 15:21:24 +00:00

mph referenced this pull request from a commit 2026-09-16 15:21:25 +00:00

Merge pull request 'fix(IW-556): keep the developer's tmux config out of the plumbing test' (#574) from IW-556 into main

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

agent-ready

  

bug

Something isn't working

  

contract

Run the tool-contract suite on this PR

  

documentation

Improvements or additions to documentation

  

duplicate

This issue or pull request already exists

  

enhancement

New feature or request

  

good first issue

Good for newcomers

  

help wanted

Extra attention is needed

  

invalid

This doesn't seem right

  

needs-human

  

needs-input

Intake asked the human a question on the issue; waiting for the answer

  

question

Further information is requested

  

wontfix

This will not be worked on

No labels

agent-ready

bug

contract

documentation

duplicate

enhancement

good first issue

help wanted

invalid

needs-human

needs-input

question

wontfix

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

2 participants

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference

iterative-works/iw-cli!574

No description provided.