Decouple fatal logging from process termination
- Dominant language
- C++
- Stars
- 221
- Forks
- 124
- Avg merge
- 1d 16h
- Merged PRs (30d)
- 21
Description
`LogLevel::kFatal` and the fatal logging macros unconditionally call `std::abort()`. `FatalHandler` runs before the abort, but it is process-global and cannot prevent termination.
This is unsafe for embedded and multi-tenant applications, where one failed Iceberg operation must not terminate the host process.
Logging should not control process lifetime. Recoverable failures should continue to use `Status`/`Result`, while termination for unrecoverable internal invariants should use a separate, explicit mechanism.
Contributor guide
No contributing guide indexed for this repository
Research direction
Start by locating the definitions and call sites for LogLevel::kFatal, the fatal logging macros, and FatalHandler. Read how recoverable failures use Status/Result and trace where std::abort() is invoked. Done should be demonstrable by logging a fatal condition without forcing process termination, while unrecoverable invariant termination remains explicit.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp
- Domain
- backend
- Issue type
- Refactor
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 30/100