Newton OS
View on Wikipedia
| Newton OS | |
|---|---|
![]() Screenshot of Newton OS 2.0 | |
| Developer | Apple Computer, Inc. |
| Written in | C++ |
| Working state | Historical |
| Source model | Closed source |
| Initial release | August 3, 1993 |
| Final release | 2.1 / 21 March 1997 |
| Supported platforms | ARM |
| Kernel type | Microkernel |
| Default user interface | GUI |
| License | Proprietary EULA |
Newton OS is a discontinued operating system for the Apple Newton PDAs produced by Apple Computer, Inc. between 1993 and 1997. It was written entirely in C++ and trimmed to be low power consuming and use the available memory efficiently. Many applications were pre-installed in the ROM of the Newton (making for quick start-up) to save on RAM and flash memory storage for user applications.
Features
[edit]Newton OS features many interface elements that the Macintosh system software didn't have at the time, such as drawers and the "poof" animation. An animation similar to this is found in Mac OS X, and parts of the Newton's handwriting recognition system have been implemented as Inkwell in Mac OS X.
- Sound responsive — Clicking menus and icons makes a sound; this feature was later introduced in Mac OS 8.
- Icons - Similar to the Macintosh Desktop metaphor, Newton OS uses icons to open applications.
- Tabbed documents — Similar to tabbed browsing in today's browsers and Apple's At Ease interface, documents titles appear in a small tab at the top right hand of the screen.
- Screen rotation — In Newton 2.0, the screen can be rotated to be used for drawing or word processing.
- File documents — Notes and Drawings can be categorized. E.g. Fun, Business, Personal, etc.
- Print documents — Documents on the Newton can be printed.
- Send documents — Documents can be sent to another Newton via Infrared technology or sent using the Internet by E-Mail, or faxed.
- Menus — Similar to menus seen in Mac OS, but menu titles are instead presented at the bottom of the screen in small rectangles, making them similar to buttons with attached "pop-up" menus.
Many features of the Newton are best appreciated in the context of the history of Pen computing.[1][2][3]
Software
[edit]Shortly after the Newton PDA's release in 1993, developers were not paying much attention to the new Newton OS API and were still more interested in developing for the Macintosh and Windows platforms. It was not until two years later that developers saw a potential market available to them in creating software for Newton OS. Several programs were made by third-party developers, including software to enhance the disappointing hand writing recognition technology of Newton OS 1.x.
The basic software that came with Newton OS:
- Works — A program for drawing and word processing, with typical capabilities such as: rulers, margins, page breaks, formatting, printing, spell checking and find & replace tools.
- Notes — Used for checklists, as well as both drawing and writing in the same program either with a newton keyboard or a stylus pen.
- Dates — Calendar program where you can schedule appointments and other special events.
- Names — Program for storing extensive contacts information in a flexible format.
- Formulas — Program that offers metric conversions, currency conversions, loan and mortgage calculators, etc.
- Calculator — A basic calculator with square root, percentage, MR, M+ and M− functions additional to the basic functions found on a calculator.
- Clock — A small floating window type application, known as a desktop accessory on the Macintosh. The Newton clock also includes features for an alarm, minute timer and the date.
- Book Reader — Support for displaying electronic books is built in.
Version history
[edit]| Date released | OS version |
|---|---|
| August 3, 1993 | 1.0 |
| October 30, 1993 | 1.1 |
| ? | 1.2 |
| March 4, 1994 | 1.3 |
| March 14, 1996 | 2.0 |
| March 21, 1997 | 2.1 |
Handwriting recognition
[edit]The Newton uses the CalliGrapher word-based handwriting recognition engine developed by ParaGraph International Inc, led by former Soviet scientist Stepan Pachikov.[4][5]
The earliest versions had weaknesses that resulted in bad publicity and reviews.[6][7][8][9] However, with the release of Newton PDAs based upon version 2.0 of the OS, the handwriting recognition substantially improved, partially being a product of ParaGraph and an Apple-created recognizer pair: Apple's Rosetta and Mondello. Newton's handwriting recognition, particularly the print recognizer, has been considered by many reviewers, testers, and users to be the best in the industry, even 10 years after it was introduced.[10] It was developed by Apple's Advanced Technology Group, and was described in 2012 as "the world's first genuinely usable handwriting recognition system".[11]
The Newton can recognize hand-printed text, cursive, or a mix of the two, and can also accept free-hand "Sketches", "Shapes", and "ink text". Text can also be entered by tapping with the stylus on a small on-screen pop-up QWERTY keyboard. With "Shapes", Newton can recognize that the user was attempting to draw a circle, a line, a polygon, etc., and it cleans them up into "perfect" vector representations (with modifiable control points and defined vertices) of what the user is attempting to draw. "Shapes" and "Sketches" can be scaled or deformed once drawn. "Ink text" captures the user's free-hand writing but allows it to be treated somewhat like recognized text when manipulating for later editing purposes ("ink text" supported word wrap, could be formatted to be bold, italic, etc.).[12] At any time a user can also direct the Newton to recognize selected "ink text" and turn it into recognized text (deferred recognition). A Newton Note document (or the notes attached to each contact in Names and each calendar event) can contain any mix of interleaved text, ink text, Shapes, and Sketches.[13]
NewtonScript
[edit]Newton OS runs applications written in C++, along with an interpreted, user-friendly language called NewtonScript. These applications are stored in packages.
Timeline
[edit]| Timeline of Newton models |
|---|
![]() |
See also
[edit]Notes
[edit]- A selection of PDFs of Apple's Newton manuals
- Newton FAQ
- Pen Computing's First Look at Newton OS 2.0
- The Newton Hall of Fame: People behind the Newton
- Pen Computing's Why did Apple kill the Newton?
- Pen Computing's Newton Notes column archive
- A.I. Magazine article by Yaeger on Newton HWR design, algorithms, & quality[dead link] and associated slides
- Info on Newton HWR from Apple's HWR Technical Lead
References
[edit]- ^ Notes on the (relatively unknown) History of Pen-based Computing
- ^ Annotated Bibliography in Pen Computing[permanent dead link]
- ^ Notes on the History of Pen-based Computing on YouTube
- ^ Calore, Michael (23 January 2007). "Stepan Pachikov Wants to Study Your Handwriting". WIRED. Archived from the original on 22 December 2008. Retrieved 6 February 2009.
- ^ Luckie, Douglas. "Newton Hall of Fame". Retrieved 6 February 2009.
- ^ "Classic Apple Moment: Newton Handwriting Recognition".
- ^ "The Story Behind Apple's Newton". Archived from the original on 28 August 2017. Retrieved 10 September 2017.
- ^ "Five Gadgets Ahead of Their Time Apple Newton". 25 October 2010. Archived from the original on 28 October 2010.
- ^ "Will the iPhone's Siri have its 'Egg freckles' moment?". CNN. Archived from the original on 7 April 2013. Retrieved 1 October 2012.
- ^ HWR accuracy:
- See comments in Wired's Apple Newton Just Won't Drop (four years later)
- See text under "Handwriting Recognition" in Pen Computing's First Look at Newton OS 2.0
- See "Opportunity Squandered" in Pen Computing's Why did Apple kill the Newton?
- See comments under "Software" in MacTech's MessagePad 2000 review
- Comments by Pen Computing's editor
- See user testing results discussed in part 6 of this A.I. Magazine article on Newton HWR[dead link]
- MessagePad 2000 review at Small Dog Electronics
- See comments under "Note-taking" in MessagePad 2000 review at "The History and Macintosh Society"
- What's Right With The Newton: HWR Archived September 8, 2006, at archive.today
- Assessment by Apple developer Archived July 17, 2012, at the Wayback Machine
- Prof. Warden's MessagePad 130 review Archived December 31, 2006, at the Wayback Machine
- ^ Smykil, Jeff (June 2005). "In the Loop with Larry Yaeger: part I". Ars Technica. Retrieved 27 February 2012.
Also at Apple, in the Advanced Technology Group, [Larry Yaeger] was Technical Lead in the development of the neural network-based handprint recognition system, the world's first genuinely usable handwriting recognition system, showcased in second generation Newton PDAs and Mac OS X's "Inkwell."
- ^ Pen Computing's First Look at Newton OS 2.0
- ^ Read about HWR, ink text, Sketches, & Shapes in Apple's MessagePad Handbook available in Apple's Newton Manuals Archived April 6, 2008, at the Wayback Machine collection
External links
[edit]Newton OS
View on GrokipediaHistory and Development
Origins and Conception
The Newton OS project originated in 1987 as a skunkworks initiative within Apple, spearheaded by then-engineering chief Jean-Louis Gassée to retain key talent Steve Sakoman, who envisioned a revolutionary personal digital assistant (PDA).[2] Inspired by the futuristic concepts in Apple's 1987 Knowledge Navigator video—which depicted a clamshell tablet with voice-activated intelligence and seamless information access—and research from Xerox PARC on graphical interfaces and portable computing, the project aimed to create a device that transcended traditional computers by integrating advanced user interactions.[2][4] Gassée approved the effort, establishing a semi-autonomous team in an abandoned Cupertino warehouse to foster innovation away from Apple's core hardware divisions.[5] The core conception focused on a portable, pen-based device roughly the size of a folded A4 sheet, priced comparably to a desktop computer, with handwriting recognition as a foundational technology to enable natural input without keyboards.[2] Specific goals included unifying scheduling, note-taking, and communication functions into a single, intelligent system that could wirelessly share data, reflecting a vision for a "knowledge navigator" that anticipated modern PDAs.[5] Early planning emphasized modular design principles.[2] Gassée led the project until his resignation from Apple in March 1990 amid internal conflicts over pricing strategy, after which Sakoman left Apple later that year to co-found Be Inc. with Gassée. Apple retained control of the Newton project under new leadership, including Michael Tchao who revitalized the effort.[2] Initial prototypes emerged from this period, testing handwriting-to-text conversion and basic PDA functionalities, with an ambitious target release date of April 1992 for the first device, codenamed Figaro.[2] This timeline underscored the project's high aspirations, though hardware challenges like processor selection would later delay commercialization.[5]Key Milestones and Releases
In early 1992, Apple formalized its Newton project as a dedicated initiative to develop a personal digital assistant platform, establishing internal teams focused on hardware and software integration.[6] This effort included a key manufacturing partnership with Sharp Electronics, which agreed to produce Newton devices in Japan and license Apple's software for its own compatible products.[6] Additional collaborations emerged with Motorola for wireless integration, as well as Cirrus Logic for custom chipsets in later Newton products.[7][8] On May 29, 1992, Apple publicly announced the Newton platform at the Consumer Electronics Show (CES) in Chicago, where CEO John Sculley unveiled prototypes and conducted the first demonstrations, including a notable demo of ordering pizza via the device's interface.[1] The announcement generated significant buzz for the platform's innovative stylus-based interaction and handwriting recognition capabilities, positioning it as a bridge between computing and consumer electronics.[9] Development faced notable delays, primarily stemming from challenges in refining the handwriting recognition system to meet performance expectations, pushing the commercial launch from an initial target of spring 1992 to mid-1993.[10] These issues drew internal scrutiny, which intensified years later following Steve Jobs' return to Apple in 1997, leading to increased pressure on the project's viability.[11] The first product, the Newton MessagePad 100, shipped on August 3, 1993, bundled with Newton OS 1.0 and featuring a monochrome display and integrated applications for notes, contacts, and scheduling.[12] Following the launch, Apple expanded ecosystem connectivity with the release of the Newton Connection Kit in 1994, enabling data synchronization between Newton devices and PowerBook laptops via serial cable, which facilitated backups and file transfers to enhance cross-device productivity.[13]Core Architecture
System Design and Components
Newton OS employed a microkernel-based architecture designed for efficiency in resource-constrained handheld devices, utilizing a 20 MHz ARM 610 processor to prioritize low power consumption and portability.[14] The microkernel served as a lightweight foundation, providing multi-tasking through tasks as concurrency units switched approximately 50 times per second, along with preemptive priority-based scheduling, ports and messages for inter-task communication, semaphores, and monitors for synchronization.[14] This modular design enabled higher-level services to operate concurrently while minimizing overhead, supporting the system's object-oriented framework written primarily in C++.[14] Memory management in Newton OS utilized a single address space model with a shared multi-level page table, accommodating up to 16 domains for isolation and protection.[14] Virtual memory was implemented via fault-driven domain managers, allowing sub-page permissions in 1 KB increments to optimize the limited RAM, which started at 640 KB of static RAM in early models like the original MessagePad and scaled to 8 MB in later devices such as the MessagePad 2100 (4 MB DRAM plus 4 MB Flash RAM).[14][15][16] Data persistence relied on the Newton Database (NDB), an object-oriented storage system using "soups"—flat, indexed collections of frames (key-value pairs)—with automatic garbage collection and transactional updates to handle dynamic content efficiently.[17] Power management was integral to extending battery life, with the OS leveraging ARM's low-power design to achieve instructions per second per watt optimization, as continuous processor operation would otherwise limit runtime to minutes.[14] Features included sleep modes that reduced activity during idle periods, enabling up to 24 hours of operation on a single charge for models like the MessagePad 2100, and dynamic resource allocation to suspend non-essential tasks.[18] Battery status monitoring via APIs like BatteryStatus provided voltage, capacity, and charge state feedback, supporting informed power conservation.[17] The file system adopted a flat, database-driven structure centered on NDB soups, eschewing traditional hierarchical folders in favor of semi-transparent persistence for objects, which facilitated seamless data portability across devices and applications through formats like packages and virtual binary objects (VBOs).[14][17] Non-volatile Flash cards provided durable storage, while SRAM cards required battery backup for data retention when powered off, with automatic saving every 30 seconds or after brief inactivity to maintain data integrity.[17][19][20] Networking capabilities were embedded in the OS stack, supporting AppleTalk for local area connectivity, serial cables with dynamic driver loading via smart protocols, and early PCMCIA modems for fax and email over phone lines.[14] Dial-in networks and IrDA infrared links enabled remote access and peer-to-peer beaming, with Type II PCMCIA slots accommodating modems like the 28.8 Kbps NovaModem for broader internet integration.[17][19]User Interface and Interaction Model
The Newton OS utilized a grayscale LCD display with resolutions varying by model (e.g., 336×240 pixels on the MessagePad 100 or 480×320 on the MessagePad 2000), some backlit in later models, optimized for stylus-based interaction on portable devices like the MessagePad series.[21][22] This interface employed a soup-based metaphor for data organization, where information was stored and retrieved as dynamic, object-oriented entries in lightweight databases called "soups," facilitating seamless integration across applications without rigid file hierarchies.[23] For instance, the built-in Notepad application exemplified this approach by treating notes as freeform entries on an infinite paper roll, allowing users to capture handwriting, sketches, or typed text in a fluid, non-linear manner.[24] Navigation and interaction centered on gesture-based controls, with tapping to select items like buttons, checkboxes, or list entries; dragging to reposition views or scroll content via universal arrows; and specialized gestures such as "scrubbing" to erase text or drawings, which triggered an animated deletion effect.[24] The overview shelf provided a multi-tasking hub, displaying thumbnails of open applications for quick switching, while the system supported concurrent operations—such as sending a fax in the background during foreground use of another app—thanks to its preemptive multithreaded architecture.[25] This allowed background tasks to continue with minimal interruption, notifying users via status slips when complete.[24] Accessibility was enhanced through visual and auditory cues, including sound feedback for stylus taps and adjustable interface elements like large, tappable buttons that accommodated finger input alongside stylus use.[24] Users could customize the experience by configuring shelves to prioritize frequently used applications, repositioning information panes for workflow efficiency, and setting preferences for input recognition to suit individual needs.[24] These features, combined with the system's reliance on handwriting as the primary input method, created an intuitive, paper-like interaction paradigm tailored to mobile productivity.[24]Key Technologies
Handwriting Recognition
The handwriting recognition system in Newton OS, part of the broader Newton Intelligence subsystem, initially relied on the CalliGrapher engine developed by ParaGraph International and licensed to Apple. This engine supported recognition of printed, cursive, and block letters, as well as mixed styles, and was designed to process stylus input in real time across multiple languages.[26][27] In early versions like Newton OS 1.x, users could train the system by writing entries in the Names database and correcting misrecognitions, allowing the engine to adapt to individual handwriting styles and vocabulary, such as personal names or specialized terms. This training process incrementally improved accuracy through dictionary augmentation and pattern learning.[28] With the release of Newton OS 2.0, Apple introduced the Rosetta engine as a dedicated print recognizer, complementing the existing cursive capabilities of CalliGrapher. Rosetta focused on block and printed characters, achieving significantly higher accuracy—often described as near-flawless for standard printed input without initial training—and extended support for additional languages while reducing dependency on user-specific dictionaries.[29] By OS 2.0, the printed recognition achieved approximately 95% accuracy on alphanumeric test sets, with user adaptation further improving results, thanks to Rosetta's neural network-based approach and adaptive algorithms that refined predictions based on corrections across applications.[29] The system also incorporated Printed Character Identification (PCI) enhancements, enabling more robust handling of isolated characters and symbols in multilingual contexts.[29] Beyond text, Newton OS featured shape recognition for interpreting diagrams, charts, and symbols drawn with the stylus, converting rough sketches into clean geometric forms like circles, rectangles, or lines. This capability integrated with the Ink Guide feature, which provided real-time visual feedback and correction, such as snapping irregular strokes to precise shapes or guiding users toward legible input zones on the screen.[24] Adaptive learning algorithms further evolved the system by analyzing user corrections over time, prioritizing frequent patterns and reducing errors in domain-specific content like technical sketches.[30] Early versions of the handwriting recognition suffered from high error rates, particularly for cursive or sloppy input, which strained the ARM processor's computational resources and led to frequent misinterpretations. This drew widespread criticism, most notably in a 1993 Doonesbury comic strip by Garry Trudeau depicting the device mangling "Catching on?" into "egg freckles," spawning a cultural meme that symbolized the technology's initial unreliability.[31][10] Despite these limitations, subsequent updates like Rosetta mitigated much of the overhead through optimized processing, though the system still demanded deliberate printing for optimal performance on the limited hardware.[28]NewtonScript Programming Language
NewtonScript is a proprietary, prototype-based object-oriented programming language developed by Apple for creating applications on the Newton OS platform.[32] It draws inspiration from the Self language, emphasizing dynamic typing where variables and expressions do not require explicit type declarations, allowing flexible data manipulation during runtime.[32] Data is stored in structures called frames, which use slots—key-value pairs—for holding properties, methods, and other objects, enabling a lightweight and extensible object model without traditional classes.[32] Key language constructs revolve around prototypes for inheritance, where objects inherit behavior and data from parent prototypes via special slots like_proto for cloning and _parent for delegation.[32] Methods are defined as functions within frame slots and invoked using message-send syntax (e.g., object:method()) or the doMethod function for dynamic dispatch, with the self pseudo-variable providing access to the receiver's context.[32] User interface elements are handled through views, which are frames representing on-screen components with attached scripts for event handling and layout.[33] Memory management is automatic via garbage collection, which activates when system memory is low to reclaim unused objects, ensuring efficient resource use on the constrained Newton hardware.[32]
The primary development environment for NewtonScript is the Newton Toolkit (NTK), an integrated application for Macintosh systems that combines code editing, UI design, and debugging tools.[33] NTK supports drag-and-drop UI building using protoViews—predefined or custom templates like protoApp or protoLabelInputLine—which developers can assemble visually and link to NewtonScript code for behavior definition.[33] Projects in NTK compile into packages deployable to Newton devices via serial or AppleTalk connections, facilitating iterative development directly from Macintosh.[33]
A significant advantage of NewtonScript is its support for hot-swapping code at runtime, allowing developers to modify methods and slots in live applications without restarting, which accelerates prototyping and testing on the device.[32] With the release of NewtonScript 2.0 in Newton OS 2.0, the language gained advanced features including closures for capturing lexical scopes in functions, namespaces for organizing code and avoiding global pollution, enhanced subclassing mechanisms, and support for native functions to improve performance.[32] These enhancements enabled more sophisticated application logic while maintaining the language's core simplicity.
For example, a simple NewtonScript snippet to parse a custom Notepad entry from a soup—representing a frame with text and metadata slots—might look like this:
func ParseNotepadEntry(entry)
begin
local text := entry.text or "";
local date := entry.date or currentDate();
if [String](/page/String)(text) then
return {summary: [String](/page/String)(text), created: date}
else
Throw(kSoupEntryError, "Invalid text slot");
end;
This function extracts and validates the text slot, defaults the date if absent, and returns a structured frame, demonstrating dynamic slot access and exception handling.[32] NewtonScript powers built-in applications like Datebook for managing event data through scripted views and storage.[32]

