ncurses: Portable Screen-Handling for Linux
Since you are clueful enough to subscribe to Linux Journal, you may already know that Linux includes a clone of the Unix screen-handling library, curses(3). You may also know that Linux's ncurses(3), like its ancestor, is a general screen-handling library that supports all popular asynchronous terminal types, including the color-ANSI and vt100-like capabilities of most Linux consoles. You've certainly run a program that used curses, vi, perhaps, or any of a dozen screen-oriented games.
You may not know that there are actually two major flavors of curses, with differences that are important for portability. You probably don't know (unless you have read ahead) that post-1.9 releases of ncurses include dramatic new features that will make them far more reliable, powerful, and useful than their predecessors.
The first curses library was hacked together at the University of California at Berkeley in about 1980 to support a screen-oriented dungeon game called rogue. It leveraged an earlier facility called termcap, the terminal capability library, which was used in the vi editor and elsewhere.
Termcap consisted of a simple text database format for describing the control sequences and capabilities of serial terminals. Before the ANSI standard hundreds of mutually-incompatible terminal types competed in the market; termcap described them in a standard way that made it possible for vi's screen-painting code to be terminal-independent.
The curses library took this one step further, by defining a relatively clean C API (applications programming interface) to hide the details of termcap-based screen painting. This facility was a terrific success, and gets almost all the credit for lifting Unix out of the line-oriented interface style inherited from its early days on ASR-33 Teletypes.
The success of curses did not go unnoticed at Bell Labs. Later System III releases and System V Release 1 included an enhanced curses library with many new features. These included:
Support for multiple screen highlights (BSD curses could only handle one “standout” highlight, usually reverse-video).
Support for line- and box-drawing using forms characters.
Recognition of function keys on input.
Color support (in later versions).
Full support for sub-windows.
Support for pads (windows of larger than screen size on which the screen or a sub-window defines a viewport).
The Bell Labs curses was upward-compatible from BSD curses, and much more powerful. However, its designers made one controversial major change; they scrapped the all-text termcap format for terminal capability descriptions, opting instead for a binary format called terminfo.
The most important reason for changing formats was that the BSD termcap database was getting so large that sequential searching took significant overhead. The terminfo format (binary capability blocks living in a bushy directory tree), by contrast, was optimized for fast lookup and fast loading.
In doing changing to a binary format, however, Bell Labs curses gave up two valuable features: extensibility and the ability to edit terminal descriptions with standard text tools. The termcap routines had never actually cared what the capabilities meant to the program using them, whether it was curses or something else. Thus, it was easy to add new capabilities and new interpretations; in fact, later BSD releases used the same code for printer- and modem-capability databases. In terminfo, by contrast, additions to the database format required a recompile of the library.
AT&T's no-source-code policy meant that Bell Labs curses itself never got a chance to outcompete the BSD version in the hacker community. Worse, the tradeoffs in the terminfo format meant that the largely BSD-centric hacker culture had a passable excuse to denigrate the major new capabilities in Bell Labs curses.
For more than five years, then, freeware authors writing screen-oriented programs were in the vexing situation of having to design around the older, less-capable BSD interface even on machines that supported the Bell Labs curses.
Around 1982, noted hacker Pavel Curtis (formerly of Xerox PARC, now perhaps best known for his MOO project) attacked this problem head-on by starting work on a freeware clone of Bell Labs curses. This package was part of Pavel's personal toolkit; it was not widely distributed, or even very well known, until Zeyd Ben-Halim, firstname.lastname@example.org, took over the development effort in late 1991. I got involved in late 1993 to support a screen-oriented multi-user Unix BBS I was developing.
The early ncurses versions had some serious drawbacks. The biggest problem was that a good many important Bell Labs curses capabilities were absent, leaving annoying and unpredictable gaps in the API. The documentation was poor and not suited to on-line browsing. Very little systematic compatibility testing against Bell Labs curses had been done. Last, but not least, all versions up to 1.8.5 had serious bugs.
|Speed Up Your Web Site with Varnish||Jun 19, 2013|
|Non-Linux FOSS: libnotify, OS X Style||Jun 18, 2013|
|Containers—Not Virtual Machines—Are the Future Cloud||Jun 17, 2013|
|Lock-Free Multi-Producer Multi-Consumer Queue on Ring Buffer||Jun 12, 2013|
|Weechat, Irssi's Little Brother||Jun 11, 2013|
|One Tail Just Isn't Enough||Jun 07, 2013|
- Speed Up Your Web Site with Varnish
- Containers—Not Virtual Machines—Are the Future Cloud
- Linux Systems Administrator
- Lock-Free Multi-Producer Multi-Consumer Queue on Ring Buffer
- RSS Feeds
- Senior Perl Developer
- Technical Support Rep
- Non-Linux FOSS: libnotify, OS X Style
- UX Designer
- So when they found it hard to
11 min 44 sec ago
33 min 55 sec ago
- Reply to comment | Linux Journal
56 min 14 sec ago
- Android has been dominating
1 hour 46 sec ago
- It is quiet helping
3 hours 46 min ago
4 hours 3 min ago
- Reachli - Amplifying your
5 hours 19 min ago
6 hours 8 min ago
- good point!
6 hours 11 min ago
- Varnish works!
6 hours 20 min ago
Free Webinar: Hadoop
How to Build an Optimal Hadoop Cluster to Store and Maintain Unlimited Amounts of Data Using Microservers
Realizing the promise of Apache® Hadoop® requires the effective deployment of compute, memory, storage and networking to achieve optimal results. With its flexibility and multitude of options, it is easy to over or under provision the server infrastructure, resulting in poor performance and high TCO. Join us for an in depth, technical discussion with industry experts from leading Hadoop and server companies who will provide insights into the key considerations for designing and deploying an optimal Hadoop cluster.
Some of key questions to be discussed are:
- What is the “typical” Hadoop cluster and what should be installed on the different machine types?
- Why should you consider the typical workload patterns when making your hardware decisions?
- Are all microservers created equal for Hadoop deployments?
- How do I plan for expansion if I require more compute, memory, storage or networking?