Pipecut - slides from presentation at MeetBSD California 2014
- 2. Note: About these slides • Original slides from the presentation follow – The slides are light on content, because I present without reading the slides. You probably want to watch the video instead of reading these. The slides were originally made in Inkscape / Jessyink • Past the end, I’ve included a few additional slides with details about pipecut that were demonstrated interactively
- 12. (Switch to a shell and show example commands)
- 14. Additional information about pipecut • The previous slide was the end of the deck used for the presentation at MeetBSD, it was followed by a demonstration of the tool. • The code for pipecut will be hosted here: • https://code.google.com/p/pipecut/ • Project homepage: • http://www.pipecut.org/
- 15. What is pipecut? • A tool for developing Unix pipeline command lines more quickly and easily • An interactive, vi-like curses interface that lets you see the data and edit the command pipeline on the same screen (no need to run and quit more/less over and over again) • A way to save useful commands and recall them later, or share them with others
- 16. How does pipecut make commandline development faster & easier? 1/2 • Single-keypress hotkeys for adding common Unix commands • Real-time highlighting of matches while typing regular expressions • The ability to see the data transformation applied instantly as you add a ‘blade’ on the command line. See just one screenful, or process the entire data file • The ability to browse the state of the data at any stage by using the cursor keys (left and right) to ‘scrub’ through the ‘blades’
- 17. How does pipecut make commandline development faster & easier? 2/2 • Works with every existing Unix command by using pipe(2) and treating it as a black-box – just works better with built-in commands • Optionally, start with your familiar environment on the command line, and run history | pipecut to take your pipeline into the pipecut environment • Give a ‘toolset’ and name and description, and it gets saved in your SQLite toolbox for use again later • With one keypress, generate a shell script that executes your toolset.
- 18. What else will pipecut do in the future? 1/3 • Generate code that runs your toolset in other target languages. • Generate its own regression tests by comparing internal/external command implementations and generated code results • Provide accelerated implementations for command combinations such as ‘sort | uniq –c | sort –nr’ where a counted hashmap can be more efficient than standalone commands without context of how the data will be processed next (executing in pipecut, or in generated code)
- 19. What else will pipecut do in the future? 2/3 • Add interactive support for more utilities’ arguments (Regex, field-references, byte-references) • Static analysis of regular expressions (e.g. highlight syntactical and semantic errors) • Optimization of pipeline display to users’ tastes (e.g. egrep –v ‘foo’ | egrep –v ‘bar’ vs egrep –v ‘foo|bar’ ) • Remappable keyboard commands
- 20. What else will pipecut do in the future? 3/3 • Interactive command help • Possess a deep understanding of command line options on each utility • Allow user to force Regex highlighting mode while typing black-box commands
- 21. Pipecut • The code for pipecut will be hosted here: • https://code.google.com/p/pipecut/ • Project homepage: • http://www.pipecut.org/ • Target for first release – Dec 24th, 2014