Sessions.factories() silently discards failing Docker session factories

[shaaravraghu]

Type
Bug

Severity
Medium

Modules

• spawn-docker

Evidence

• spawn-docker/src/main/java/build/spawn/docker/Sessions.java:64-74
• Provider creation and DI injection are wrapped in a try/catch.
• Any exception causes return null, followed by .filter(Objects::nonNull).

Why this matters

• Broken service registration, wiring errors, or initialization problems disappear without any logging.
• Callers then only see Optional.empty() from createSession(...), which looks like "no Docker available" rather than "a factory crashed".

Expected behavior

• A failing factory should either be logged with context or surfaced explicitly.

Actual behavior

• Factory failures are swallowed silently.
• Diagnostics for session discovery become much harder than necessary.

Suggested reproduction

1. Introduce a session factory whose constructor or injection path throws.
2. Call Sessions.createSession(...).
3. Observe that the broken factory disappears silently from discovery.

Suggested fix

• Log the exception together with the provider class name.
• Consider continuing after logging, but do not silently erase the failure.

Suggested tests

• Add a fake factory in test scope that throws during injection and assert that the failure is observable.

Issue body ready to paste
Sessions.factories() catches provider/injection exceptions and returns null, which is then filtered out. This makes broken Docker session factories vanish silently and turns real wiring failures into an unhelpful Optional.empty() from createSession(...).`