Container Manager 1.0.8

Stacks of stacks.

New

Compose files that actually run. Imports now handle the things that
previously stopped them dead:

• ${VAR} references become fields on the create form, prefilled from a .env
  beside the compose file — or loaded any time with Import .env….
  Password-like names are masked automatically.
• command: is carried through instead of dropped, so images that need one
  (Fleet, Redis with flags) import intact.
• platform: is honoured, so linux/amd64-only images run under emulation on
  Apple silicon.
• Whatever can't be carried over is now listed with the reason and the
  consequence, rather than just a list of key names.

Stacks are editable. Add, replace, or remove a service on an existing stack.
Replacing is the way to repair one that failed or needs different settings — it
keeps the stack's network, labels, and web URL, and leaves your data volumes
alone.

Stacks remember how they were built. The definition and the values you
entered are saved with the stack, so a stack missing a service offers
Re-create to rebuild it exactly as defined — no retyping image, command, env,
ports, or platform.

Services wait for their dependencies. Each service is waited on until it
accepts connections before the next one starts, so a database still initialising
no longer causes the app that depends on it to fail on startup.

Environment files for images and containers. The Build Image sheet gains an
Environment section with .env import; values are passed as build args and
baked in as image defaults. The container create sheet gets the same import.

Better visibility into stacks. A stack now shows its network (name, mode,
subnet), the volumes its services mount and where, and a Log of how it
was built — including what the import skipped and any failure — so you can work
out what still needs doing.

Open a terminal into a stack service. Right-click any service for an in-app
terminal or Terminal.app; the shell sees the stack's volumes mounted.

Group volumes by label. The Volumes list groups into collapsible sections.
Volumes created for a stack are labelled automatically, and you can label any
others yourself — including several at once.

Stacks are checked before anything is created. Missing host paths, mount
paths left empty by an unset variable, invalid ports, and missing images are all
reported together up front, instead of building half a stack and failing on the
last service.

Fixed

• The app froze while creating a stack. Image downloads and container
  creation ran on the main thread, so the window couldn't repaint or scroll —
  progress was invisible and it looked hung. That work now runs in the
  background, and the progress log stays put and readable throughout.
• A failed create didn't say what happened. The sheet showed "Cancel" next to
  a re-armed "Create" as though nothing had occurred, while services created
  before the failure were in fact running. It now reports how many services were
  created, why it stopped, and offers Close or Retry.
• Pasting a compose file into the Build Image sheet produced a cryptic
  unknown instruction: services:. It's now recognised, pointing you at
  Stacks ▸ New Stack ▸ Import Template… instead.

Known limitations

Stack services reach each other by IP rather than hostname, and a service's
environment is fixed when it's created — so recreating a dependency leaves its
dependants pointing at the old address. See the
stack definitions guide for the full list.