todo.md

bless TODO - Missing less Features

Navigation

• F - Follow mode (like tail -f, auto-scroll as file grows)
• r / R - Repaint/refresh screen
• z - Forward one window (with optional count)
• w - Backward one window
• e - Forward one line (alias for j)
• y - Backward one line (alias for k)

Searching

• ? - Search backward
• &pattern - Show only lines matching pattern (filter mode)
• ESC-u - Toggle search highlighting

Marks

• m + letter - Set a mark at current position
• ' + letter - Jump to a mark
• '' - Return to previous position before last jump

File Commands

• :d - Remove current file from buffer list
• :x - Go to first file in buffer list

Options (toggle with -)

• -i - Ignore case in search
• -I - Ignore case even when pattern has uppercase
• -N - Toggle line numbers
• -S - Toggle line wrap (chop long lines)

Other

• v - Open current file in $EDITOR
• h - Help screen (less standard, we have F1)

---

Command-Line Arguments

Display Options

• -N / --LINE-NUMBERS - Show line numbers
• -S / --chop-long-lines - Don't wrap, truncate long lines
• -R / --RAW-CONTROL-CHARS - Interpret ANSI color escape codes
• -X / --no-init - Don't clear screen on exit
• -x N / --tabs=N - Set tab width

Search Options

• -i / --ignore-case - Ignore case in searches
• -I / --IGNORE-CASE - Ignore case even if pattern has uppercase
• -p pattern - Start at first occurrence of pattern
• +/pattern - Same as -p

Behavior Options

• -F / --quit-if-one-screen - Quit if content fits on one screen
• -e / --quit-at-eof - Quit at second EOF
• -E / --QUIT-AT-EOF - Quit at first EOF
• -q / --quiet / --silent - No terminal bell
• +F - Start in follow mode
• +G - Start at end of file

Prompt Options

• -M / --LONG-PROMPT - Verbose prompt with line numbers/percentage
• -m / --long-prompt - Medium verbosity prompt

---

Architecture: Large File Support

Currently bless reads entire files into memory (os.ReadFile), which fails for large files (100M+ lines). Need to implement lazy loading like less.

Line Index

• Build index of byte offsets for line starts
• Use sparse index (every Nth line) for memory efficiency
• Enable O(log n) jumping to any line number
• Incremental index building (don't block on startup)

Lazy Loading

• Memory-map files instead of reading entirely
• Load only visible lines + buffer region
• LRU cache for recently viewed chunks
• Re-read from disk when scrolling to new regions

Viewport-Based Rendering

• Decouple viewport from full document storage
• Only materialize text for visible lines
• Handle word-wrap calculation without full content

Stdin / Pipe Handling

• Buffer stdin to temp file for random access
• Support seeking in piped content
• Size limits / warnings for huge streams

Performance Targets

• Open 1GB+ files instantly (index in background)
• Smooth scrolling regardless of file size
• Memory usage proportional to viewport, not file size

---

Priority Features

High impact features to implement first:

Keybindings

1. Follow mode (F) - Essential for log files (DONE)
2. Backward search (?) - Common navigation pattern
3. Marks (m/') - Power user feature for large files
4. Case-insensitive search toggle (-i) - Very frequently needed
5. Filter mode (&) - Useful for log analysis

Command-Line Arguments

1. -N - Line numbers (common use case)
2. -S - Chop long lines (essential for wide content)
3. -i - Case-insensitive search (very common)
4. -F - Quit if one screen (used in scripts/pipes)
5. +F - Start in follow mode (log file monitoring)
6. -R - Raw ANSI codes (colored output from commands)