IKE Working-Set Leases
IKE Working-Set Leases 6-SNAPSHOT
- Documentation Overview
- Reports Project Information 12
IKE Working-Set Leases
The working-set lease machinery of IKE’s multi-machine development fabric: one writer at a time per working set in a Syncthing-synchronized development folder, with the synced lease record as both grant and bus. Delivered in full under IKE-Network/ike-issues#1002[1] and its descendants; canonical design in the dev-working-set-lease topic of the lab corpus.
Two Maven modules, one reactor (network.ike:ike-lease):
| Module | What it is |
|---|---|
ike-lease-core |
The canonical lease protocol — verbs, epoch fencing, TTL staleness, reconcile-on-read, confirmed acquisition, ref stamps — plus git-state materialization for synced bare trees, root ref alignment to the holder’s stamps, the reconciliation daemon, and the headless CLIs. IntelliJ-free by construction. |
ike-lease-plugin |
The IntelliJ host: opening a project is the acquisition gesture (confirmed against the sync-propagation race), materialization and alignment run right after it, a watcher renews every open project, and a pre-push guard keeps sibling working sets off remote remotes. |
~/ike-dev/scripts/lease.sh is a thin wrapper that execs the core’s CLI, so the fence, the operator, and the IDE all run the same single implementation. Installing the plugin zip is what provisions the command line — the wrapper globs the installed lib/ for the core jar.
Deliberately outside the release cascade
This repository releases manually (ike:release-draft / ike:release-publish), publishes to the internal Nexus and GitHub — never Maven Central — and is not a node of the foundation release cascade: the lease layer sits below the build system, and per-machine installs go through setup-machine.sh rather than artifact consumption. The full rationale, protocol description, and operating guide live in the repository README[2].