Repository navigation
min box rm and min box prune verbs over ReapBox/PruneBoxes (BCLI-004/005/006) #2834
Description
The daemon side of removing and pruning stopped boxes lands with T8 (#2395, PR #2833): the ReapBox and PruneBoxes requests, the running-box refusal, and the reaped list. No min command sends them yet.
This issue tracks the CLI verbs the BCLI spec (#1991, docs/specs/28-spec-box-cli) defines over those requests:
min box rm <box> [--force]sendsReapBoxand prints what it reaped. On a running box without--forceit reports BOX-019's refusal with exit 2, naming--force(BCLI-004).min box prune [--older-than <d>] [--parent <box|self>]sendsPruneBoxesand prints thereapedlist. Off a TTY without--yesit fails with exit 2, naming--yes(BCLI-005). With--dry-runit sums the disk the matched boxes hold (BCLI-006).--older-thancounts from when a box enteredstoppedorexited(architecture.md line 501).
It lands as a task of the BCLI plan once #1991 merges. It is listed as a residual on #2833.
Also in scope (from arch review of #2833):
- A
PruneBoxesdry-run mode on the daemon side. It lists the boxes the selectors match and the disk each holds, and reaps none (architecture.md line 501).min box prune --dry-runuses it (BCLI-006). - Amend BOX-021 in docs/specs/29-spec-box-local-first to name
--dry-run. BOX-021 omits a flag that the design names. - A disk figure in the record BOX-159 returns (folded in from BOX: return the disk a stopped or exited box holds in BOX-159 (needed by BCLI-006) #2869).
min box show <box>reports the disk that a stopped or exited box holds (BCLI-006), and the dry run above reports the same figure per box. The figure counts only what a reap frees: the retained filesystem (BOX-017), the retained output log (BOX-155) and the events stream (BOX-045). It excludes Box Volumes, becausermandpruneleave a volume in place (BVOL-008, draft docs(specs): add the Box Volumes spec (BVOL) for inbox#731 #1990). Amend BOX-159 in docs/specs/29-spec-box-local-first to name the field.
Activity
docs(specs): map every activate flag and follow arch at 7a86ec7
docs(specs): cite the BCLI residual issues and answer the review threads
docs(specs): add the min box CLI spec (BCLI) for inbox#731 (gominimal…
agent-137 commentedon Nov 4, 2026
When this adds
min box prune --dry-run(BCLI-006): the BCLI-005 refusal from T22 (#2907, PR #3008),PruneNeedsYesin crates/minimal/src/cmd/boxes/lifecycle.rs, already tells the user to run--dry-run. Until this lands, clap rejects that flag. Keep the refusal text and this flag in step, and update the test that pins the text if the wording changes.agent-137 commentedon Nov 7, 2026
The daemon half that this issue's CLI scope waited on has landed: spec 29's BOX-021 prune preview and BOX-159 disk figure, in #3078 (BOX T63).
min box rmandmin box prunemerged in #2961 and #3008. The rest,min box prune --dry-runwith the disk each box holds (BCLI-006), is BCLI T23 (#2915), which no longer waits on this issue. Left open for its owner to close.