|
 |
[2] ai.viXra.org:2607.0081 [pdf] submitted on 2026-07-28 20:41:21
Authors: Martin Placek, Rajkumar Buyya
Comments: 53 Pages. Technical Report, GRIDS-TR- 2006-11, Grid Computing and Distributed Systems Laboratory, The University of Melbourne, Australia, July 3, 2006.
This paper presents a taxonomy of key topics effecting research and development of distributed storage systems. The taxonomy finds distributed storage systems to offer a wide array of functionality, employ architectures with varying degrees of entralisation and operate a ross environments with varying trust and scalability. Furthermore, taxonomies on autonomi management, federation, onsistency and routing provide an insight into challenges faced by distributed storage systems and the research to over ome them. The paper continues by providing a survey of distributed storage systems which exemplify topics overed in the taxonomy.The selection of surveyed systems overs a variety of storage systems, exposing the reader to an array of different problems and solutions employed to over come these challenges. For each surveyed system we address the underlying operational behaviour, leading into the architecture and algorithms employed in the design and development of the system. Our survey overs systems from the past and present concluding with a discussion on the evolution of distributed storage systems and possible future work.
Category: Data Structures and Algorithms
[1] ai.viXra.org:2607.0013 [pdf] submitted on 2026-07-08 11:50:28
Authors: Brent Hartshorn
Comments: 8 Pages.
The web’s active-content model rests on JavaScript and, beneath it, on WebAssembly — a stack virtual machine that, in current browsers, must cross a JavaScript boundary for every access to the document. We report a web browser built from the ground up on the minipy/ShivyCX substrate, in which a page’s active content is either Python, executed by an embedded interpreter against a live document, or native machine code compiled ahead of time from an rpython script and invoked through ctypes—with no virtual machine, no JavaScript wrapper layer,and no third-party browser engine. Concretely: script type="python" runs against a shared document/console/window; script type="rpython" is compiled (py2c → gcc -O2) to a cached per-page shared object callable by ctypes, while the same source still runs on stock CPython; a canvas is drawn by a JIT-compiled native fragment shader with time and pointer uniforms at one native call per frame; and JavaScript runs by translation onto the same interpreter and document, not a second VM. Nothing from Blink, WebKit, Gecko, Servo, orNetSurf is used; pages render through a small software widget layer into a Wayland frame buffer. We position the result as a systems alternative to Redox OS (which ports NetSurf or Servo) and ChromeOS (which is Chrome/V8/Blink), and as a descendant of TempleOS’s ground-up philosophy— arguing that each of these optimizes within the JavaScript/engine boundary, whereas co-designing the page-execution model lets us move it.
Category: Data Structures and Algorithms