Speedy PaSh shell compiler finds new home at the Linux Foundation

Speedy PaSh shell compiler finds new home at the Linux Foundation

Sep 22, 2021 — by Eric Brown

— 100 views

The Linux Foundation has adopted the PaSh Project, which offers a Linux-based compiler for automatically parallelizing POSIX shell scripts to optimize programs and speed execution, especially on multiprocessor systems.

The PaSh compiler, which is designed for parallelizing and thereby speeding up POSIX shell scripts, has a new home with the Linux Foundation. PaSh appears to be a major advance for a shell script language that has been around for almost a quarter century and is used across the technology spectrum in utility programs such as Bash.

PaSh is designed to “improve upon and accelerate the execution of shell scripts in the face of new web crawling, indexing and natural language processing changes,” stated Nikos Vasilakis, Technical Steering Committee chair and MIT researcher.

The PaSh Project is supported by MIT, Rice University, Stevens Institute of Technology, and the University of Pennsylvania. The software is based on techniques that have gathered several awards, including a best paper award at the 16th edition of the EuroSys conference and a graduate research award from the Association of Computing Machinery for U of P’s Konstantinos Kallas, Technical Steering Committee member for the PaSh Project.

“PaSh promises a speed boost to shell users of all stripes,” stated Stevens Institute Assistant Professor and PaSh Technical Steering Committee member Michael Greenberg. On multiprocessor computers, PaSh “can execute tasks like web crawling and indexing, COVID-19 related analyses, natural-language processing, and other workloads in a fraction of their original time with minimal-to-zero developer effort,” says the announcement.

PaSh is defined as a “light-touch data-parallel shell processing” compiler that provides “source-to-source” parallelizing. The Linux-based compiler uses a POSIX shell script as an input to create a new program that is significantly faster. The source-to-source nature of the software allows the optimized shell script to be inspected and executed using the same tools, in the same environment, and with the same data as the original script. There is also a small runtime library and associated annotations on programs commonly used in shell scripts, which provides PaSh with high-performance primitives.

“As software development evolves to address machine learning, containerization, artificial intelligence and more, PaSh shows up to support developers and data scientists who need more out of their scripting tools,” stated Mike Dolan, GM and SVP Projects at the Linux Foundation.

Further information

PaSh is available for free download on its GitHub page. More information may be found at the PaSh website.