Improve bin/dev kill process management with better error handling

[justin808]

Improve bin/dev kill Process Management

Background

In the react_on_rails-demos repository, we recently implemented improved process management for watch processes with proper error handling (see PR #42). This same improvement should be applied to the bin/dev kill script in react_on_rails.

Problem

The current process killing logic likely has similar issues to what we found in the demos repo:

• Counting processes as "killed" even when Process.kill fails due to permission errors (EPERM)
• Not distinguishing between successful kills, already-stopped processes, and permission-denied errors
• Potentially confusing user feedback about what actually happened

Solution

Apply the same improved error handling pattern that was implemented in react_on_rails-demos:

Current Pattern (Incorrect)

begin
  Process.kill('TERM', pid)
rescue Errno::ESRCH, Errno::EPERM
  # Treats both cases the same
end
killed_count += 1  # Always increments, even on EPERM!

Improved Pattern

begin
  Process.kill('TERM', pid)
  killed_count += 1
rescue Errno::ESRCH
  # Process already stopped - treat as success
  puts "Process #{pid} - process no longer exists"
  killed_count += 1
rescue Errno::EPERM
  # Permission denied - do NOT count as success
  puts "⚠️  Process #{pid} - permission denied (process owned by another user)"
end

Key Improvements

1. Accurate counting: Only increment killed_count when kill succeeds or process is already gone (ESRCH)
2. Clear feedback: Distinguish between:
   • ✅ Successfully killed
   • ✅ Already stopped (ESRCH - treat as success)
   • ⚠️ Permission denied (EPERM - NOT a success)
3. Better UX: Users know exactly what happened with each process

Reference Implementation

See the complete implementation in react_on_rails-demos:

• File: lib/demo_scripts/gem_swapper.rb lines 95-118
• PR: Add watch mode process management for swap-deps (#37) react_on_rails-demos#42
• Commit: shakacode/react_on_rails-demos@04702fc

Testing Checklist

When implementing this fix:

• Test killing processes you own (should succeed)
• Test with already-stopped processes (should handle ESRCH gracefully)
• Test with processes owned by other users (should show EPERM warning and not count as success)
• Verify final count matches actual successful kills

Related

This is part of broader improvements to process management that were implemented in the demos repository for issue #37.