1 A Hacker's Guide to NCURSES
3 A Hacker's Guide to NCURSES
8 * Objective of the Package
10 + How to Design Extensions
11 * Portability and Configuration
12 * Documentation Conventions
14 * A Tour of the Ncurses Library
19 + Output and Screen Updating
20 * The Forms and Menu Libraries
21 * A Tour of the Terminfo Compiler
22 + Translation of Non-use Capabilities
23 + Use Capability Resolution
24 + Source-Form Translation
26 * Style Tips for Developers
31 This document is a hacker's tour of the ncurses library and utilities.
32 It discusses design philosophy, implementation methods, and the
33 conventions used for coding and documentation. It is recommended
34 reading for anyone who is interested in porting, extending or
35 improving the package.
37 Objective of the Package
39 The objective of the ncurses package is to provide a free software API
40 for character-cell terminals and terminal emulators with the following
42 * Source-compatible with historical curses implementations
43 (including the original BSD curses and System V curses.
44 * Conformant with the XSI Curses standard issued as part of XPG4 by
46 * High-quality -- stable and reliable code, wide portability, good
47 packaging, superior documentation.
48 * Featureful -- should eliminate as much of the drudgery of C
49 interface programming as possible, freeing programmers to think at
50 a higher level of design.
52 These objectives are in priority order. So, for example, source
53 compatibility with older version must trump featurefulness -- we
54 cannot add features if it means breaking the portion of the API
55 corresponding to historical curses versions.
59 We used System V curses as a model, reverse-engineering their API, in
60 order to fulfill the first two objectives.
62 System V curses implementations can support BSD curses programs with
63 just a recompilation, so by capturing the System V API we also capture
66 More importantly for the future, the XSI Curses standard issued by
67 X/Open is explicitly and closely modeled on System V. So conformance
68 with System V took us most of the way to base-level XSI conformance.
70 How to Design Extensions
72 The third objective (standards conformance) requires that it be easy
73 to condition source code using ncurses so that the absence of
74 nonstandard extensions does not break the code.
76 Accordingly, we have a policy of associating with each nonstandard
77 extension a feature macro, so that ncurses client code can use this
78 macro to condition in or out the code that requires the ncurses
81 For example, there is a macro NCURSES_MOUSE_VERSION which XSI Curses
82 does not define, but which is defined in the ncurses library header.
83 You can use this to condition the calls to the mouse API calls.
85 Portability and Configuration
87 Code written for ncurses may assume an ANSI-standard C compiler and
88 POSIX-compatible OS interface. It may also assume the presence of a
89 System-V-compatible select(2) call.
91 We encourage (but do not require) developers to make the code friendly
92 to less-capable UNIX environments wherever possible.
94 We encourage developers to support OS-specific optimizations and
95 methods not available under POSIX/ANSI, provided only that:
96 * All such code is properly conditioned so the build process does
97 not attempt to compile it under a plain ANSI/POSIX environment.
98 * Adding such implementation methods does not introduce
99 incompatibilities in the ncurses API between platforms.
101 We use GNU autoconf(1) as a tool to deal with portability issues. The
102 right way to leverage an OS-specific feature is to modify the autoconf
103 specification files (configure.in and aclocal.m4) to set up a new
104 feature macro, which you then use to condition your code.
106 Documentation Conventions
108 There are three kinds of documentation associated with this package.
109 Each has a different preferred format:
110 * Package-internal files (README, INSTALL, TO-DO etc.)
112 * Everything else (i.e., narrative documentation).
114 Our conventions are simple:
115 1. Maintain package-internal files in plain text. The expected viewer
116 for them is more(1) or an editor window; there is no point in
118 2. Mark up manual pages in the man macros. These have to be viewable
119 through traditional man(1) programs.
120 3. Write everything else in HTML.
122 When in doubt, HTMLize a master and use lynx(1) to generate plain
123 ASCII (as we do for the announcement document).
125 The reason for choosing HTML is that it is (a) well-adapted for
126 on-line browsing through viewers that are everywhere; (b) more easily
127 readable as plain text than most other mark-ups, if you do not have a
128 viewer; and (c) carries enough information that you can generate a
129 nice-looking printed version from it. Also, of course, it make
130 exporting things like the announcement document to WWW pretty trivial.
134 The reporting address for bugs is bug-ncurses@gnu.org. This is a
135 majordomo list; to join, write to bug-ncurses-request@gnu.org with a
136 message containing the line:
137 subscribe <name>@<host.domain>
139 The ncurses code is maintained by a small group of volunteers. While
140 we try our best to fix bugs promptly, we simply do not have a lot of
141 hours to spend on elementary hand-holding. We rely on intelligent
142 cooperation from our users. If you think you have found a bug in
143 ncurses, there are some steps you can take before contacting us that
144 will help get the bug fixed quickly.
146 In order to use our bug-fixing time efficiently, we put people who
147 show us they have taken these steps at the head of our queue. This
148 means that if you do not, you will probably end up at the tail end and
149 have to wait a while.
150 1. Develop a recipe to reproduce the bug.
151 Bugs we can reproduce are likely to be fixed very quickly, often
152 within days. The most effective single thing you can do to get a
153 quick fix is develop a way we can duplicate the bad behavior --
154 ideally, by giving us source for a small, portable test program
155 that breaks the library. (Even better is a keystroke recipe using
156 one of the test programs provided with the distribution.)
157 2. Try to reproduce the bug on a different terminal type.
158 In our experience, most of the behaviors people report as library
159 bugs are actually due to subtle problems in terminal descriptions.
160 This is especially likely to be true if you are using a
161 traditional asynchronous terminal or PC-based terminal emulator,
162 rather than xterm or a UNIX console entry.
163 It is therefore extremely helpful if you can tell us whether or
164 not your problem reproduces on other terminal types. Usually you
165 will have both a console type and xterm available; please tell us
166 whether or not your bug reproduces on both.
167 If you have xterm available, it is also good to collect xterm
168 reports for different window sizes. This is especially true if you
169 normally use an unusual xterm window size -- a surprising number
170 of the bugs we have seen are either triggered or masked by these.
171 3. Generate and examine a trace file for the broken behavior.
172 Recompile your program with the debugging versions of the
173 libraries. Insert a trace() call with the argument set to
174 TRACE_UPDATE. (See "Writing Programs with NCURSES" for details on
175 trace levels.) Reproduce your bug, then look at the trace file to
176 see what the library was actually doing.
177 Another frequent cause of apparent bugs is application coding
178 errors that cause the wrong things to be put on the virtual
179 screen. Looking at the virtual-screen dumps in the trace file will
180 tell you immediately if this is happening, and save you from the
181 possible embarrassment of being told that the bug is in your code
182 and is your problem rather than ours.
183 If the virtual-screen dumps look correct but the bug persists, it
184 is possible to crank up the trace level to give more and more
185 information about the library's update actions and the control
186 sequences it issues to perform them. The test directory of the
187 distribution contains a tool for digesting these logs to make them
188 less tedious to wade through.
189 Often you will find terminfo problems at this stage by noticing
190 that the escape sequences put out for various capabilities are
191 wrong. If not, you are likely to learn enough to be able to
192 characterize any bug in the screen-update logic quite exactly.
193 4. Report details and symptoms, not just interpretations.
194 If you do the preceding two steps, it is very likely that you will
195 discover the nature of the problem yourself and be able to send us
196 a fix. This will create happy feelings all around and earn you
197 good karma for the first time you run into a bug you really cannot
198 characterize and fix yourself.
199 If you are still stuck, at least you will know what to tell us.
200 Remember, we need details. If you guess about what is safe to
201 leave out, you are too likely to be wrong.
202 If your bug produces a bad update, include a trace file. Try to
203 make the trace at the least voluminous level that pins down the
204 bug. Logs that have been through tracemunch are OK, it does not
205 throw away any information (actually they are better than
206 un-munched ones because they are easier to read).
207 If your bug produces a core-dump, please include a symbolic stack
208 trace generated by gdb(1) or your local equivalent.
209 Tell us about every terminal on which you have reproduced the bug
210 -- and every terminal on which you cannot. Ideally, send us
211 terminfo sources for all of these (yours might differ from ours).
212 Include your ncurses version and your OS/machine type, of course!
213 You can find your ncurses version in the curses.h file.
215 If your problem smells like a logic error or in cursor movement or
216 scrolling or a bad capability, there are a couple of tiny test frames
217 for the library algorithms in the progs directory that may help you
218 isolate it. These are not part of the normal build, but do have their
219 own make productions.
221 The most important of these is mvcur, a test frame for the
222 cursor-movement optimization code. With this program, you can see
223 directly what control sequences will be emitted for any given cursor
224 movement or scroll/insert/delete operations. If you think you have got
225 a bad capability identified, you can disable it and test again. The
226 program is command-driven and has on-line help.
228 If you think the vertical-scroll optimization is broken, or just want
229 to understand how it works better, build hashmap and read the header
230 comments of hardscroll.c and hashmap.c; then try it out. You can also
231 test the hardware-scrolling optimization separately with hardscroll.
233 A Tour of the Ncurses Library
237 Most of the library is superstructure -- fairly trivial convenience
238 interfaces to a small set of basic functions and data structures used
239 to manipulate the virtual screen (in particular, none of this code
240 does any I/O except through calls to more fundamental modules
241 described below). The files
243 lib_addch.c lib_bkgd.c lib_box.c lib_chgat.c lib_clear.c
244 lib_clearok.c lib_clrbot.c lib_clreol.c lib_colorset.c lib_data.c
245 lib_delch.c lib_delwin.c lib_echo.c lib_erase.c lib_gen.c
246 lib_getstr.c lib_hline.c lib_immedok.c lib_inchstr.c lib_insch.c
247 lib_insdel.c lib_insstr.c lib_instr.c lib_isendwin.c lib_keyname.c
248 lib_leaveok.c lib_move.c lib_mvwin.c lib_overlay.c lib_pad.c
249 lib_printw.c lib_redrawln.c lib_scanw.c lib_screen.c lib_scroll.c
250 lib_scrollok.c lib_scrreg.c lib_set_term.c lib_slk.c
251 lib_slkatr_set.c lib_slkatrof.c lib_slkatron.c lib_slkatrset.c
252 lib_slkattr.c lib_slkclear.c lib_slkcolor.c lib_slkinit.c
253 lib_slklab.c lib_slkrefr.c lib_slkset.c lib_slktouch.c lib_touch.c
254 lib_unctrl.c lib_vline.c lib_wattroff.c lib_wattron.c lib_window.c
256 are all in this category. They are very unlikely to need change,
257 barring bugs or some fundamental reorganization in the underlying data
260 These files are used only for debugging support:
262 lib_trace.c lib_traceatr.c lib_tracebits.c lib_tracechr.c
263 lib_tracedmp.c lib_tracemse.c trace_buf.c
265 It is rather unlikely you will ever need to change these, unless you
266 want to introduce a new debug trace level for some reason.
268 There is another group of files that do direct I/O via tputs(),
269 computations on the terminal capabilities, or queries to the OS
270 environment, but nevertheless have only fairly low complexity. These
273 lib_acs.c lib_beep.c lib_color.c lib_endwin.c lib_initscr.c
274 lib_longname.c lib_newterm.c lib_options.c lib_termcap.c lib_ti.c
275 lib_tparm.c lib_tputs.c lib_vidattr.c read_entry.c.
277 They are likely to need revision only if ncurses is being ported to an
278 environment without an underlying terminfo capability representation.
280 These files have serious hooks into the tty driver and signal
283 lib_kernel.c lib_baudrate.c lib_raw.c lib_tstp.c lib_twait.c
285 If you run into porting snafus moving the package to another UNIX, the
286 problem is likely to be in one of these files. The file lib_print.c
287 uses sleep(2) and also falls in this category.
289 Almost all of the real work is done in the files
291 hardscroll.c hashmap.c lib_addch.c lib_doupdate.c lib_getch.c
292 lib_mouse.c lib_mvcur.c lib_refresh.c lib_setup.c lib_vidattr.c
294 Most of the algorithmic complexity in the library lives in these
295 files. If there is a real bug in ncurses itself, it is probably here.
296 We will tour some of these files in detail below (see The Engine
299 Finally, there is a group of files that is actually most of the
300 terminfo compiler. The reason this code lives in the ncurses library
301 is to support fallback to /etc/termcap. These files include
303 alloc_entry.c captoinfo.c comp_captab.c comp_error.c comp_hash.c
304 comp_parse.c comp_scan.c parse_entry.c read_termcap.c write_entry.c
306 We will discuss these in the compiler tour.
312 All ncurses input funnels through the function wgetch(), defined in
313 lib_getch.c. This function is tricky; it has to poll for keyboard and
314 mouse events and do a running match of incoming input against the set
315 of defined special keys.
317 The central data structure in this module is a FIFO queue, used to
318 match multiple-character input sequences against special-key
319 capabilities; also to implement pushback via ungetch().
321 The wgetch() code distinguishes between function key sequences and the
322 same sequences typed manually by doing a timed wait after each input
323 character that could lead a function key sequence. If the entire
324 sequence takes less than 1 second, it is assumed to have been
325 generated by a function key press.
327 Hackers bruised by previous encounters with variant select(2) calls
328 may find the code in lib_twait.c interesting. It deals with the
329 problem that some BSD selects do not return a reliable time-left
330 value. The function timed_wait() effectively simulates a System V
335 If the mouse interface is active, wgetch() polls for mouse events each
336 call, before it goes to the keyboard for input. It is up to
337 lib_mouse.c how the polling is accomplished; it may vary for different
340 Under xterm, however, mouse event notifications come in via the
341 keyboard input stream. They are recognized by having the kmous
342 capability as a prefix. This is kind of klugey, but trying to wire in
343 recognition of a mouse key prefix without going through the
344 function-key machinery would be just too painful, and this turns out
345 to imply having the prefix somewhere in the function-key capabilities
346 at terminal-type initialization.
348 This kluge only works because kmous is not actually used by any
349 historic terminal type or curses implementation we know of. Best guess
350 is it is a relic of some forgotten experiment in-house at Bell Labs
351 that did not leave any traces in the publicly-distributed System V
352 terminfo files. If System V or XPG4 ever gets serious about using it
353 again, this kluge may have to change.
355 Here are some more details about mouse event handling:
357 The lib_mouse() code is logically split into a lower level that
358 accepts event reports in a device-dependent format and an upper level
359 that parses mouse gestures and filters events. The mediating data
360 structure is a circular queue of event structures.
362 Functionally, the lower level's job is to pick up primitive events and
363 put them on the circular queue. This can happen in one of two ways:
364 either (a) _nc_mouse_event() detects a series of incoming mouse
365 reports and queues them, or (b) code in lib_getch.c detects the kmous
366 prefix in the keyboard input stream and calls _nc_mouse_inline to
367 queue up a series of adjacent mouse reports.
369 In either case, _nc_mouse_parse() should be called after the series is
370 accepted to parse the digested mouse reports (low-level events) into a
371 gesture (a high-level or composite event).
373 Output and Screen Updating
375 With the single exception of character echoes during a wgetnstr() call
376 (which simulates cooked-mode line editing in an ncurses window), the
377 library normally does all its output at refresh time.
379 The main job is to go from the current state of the screen (as
380 represented in the curscr window structure) to the desired new state
381 (as represented in the newscr window structure), while doing as little
384 The brains of this operation are the modules hashmap.c, hardscroll.c
385 and lib_doupdate.c; the latter two use lib_mvcur.c. Essentially, what
386 happens looks like this:
387 * The hashmap.c module tries to detect vertical motion changes
388 between the real and virtual screens. This information is
389 represented by the oldindex members in the newscr structure. These
390 are modified by vertical-motion and clear operations, and both are
391 re-initialized after each update. To this change-journalling
392 information, the hashmap code adds deductions made using a
393 modified Heckel algorithm on hash values generated from the line
395 * The hardscroll.c module computes an optimum set of scroll,
396 insertion, and deletion operations to make the indices match. It
397 calls _nc_mvcur_scrolln() in lib_mvcur.c to do those motions.
398 * Then lib_doupdate.c goes to work. Its job is to do line-by-line
399 transformations of curscr lines to newscr lines. Its main tool is
400 the routine mvcur() in lib_mvcur.c. This routine does
401 cursor-movement optimization, attempting to get from given screen
402 location A to given location B in the fewest output characters
405 If you want to work on screen optimizations, you should use the fact
406 that (in the trace-enabled version of the library) enabling the
407 TRACE_TIMES trace level causes a report to be emitted after each
408 screen update giving the elapsed time and a count of characters
409 emitted during the update. You can use this to tell when an update
410 optimization improves efficiency.
412 In the trace-enabled version of the library, it is also possible to
413 disable and re-enable various optimizations at runtime by tweaking the
414 variable _nc_optimize_enable. See the file include/curses.h.in for
415 mask values, near the end.
417 The Forms and Menu Libraries
419 The forms and menu libraries should work reliably in any environment
420 you can port ncurses to. The only portability issue anywhere in them
421 is what flavor of regular expressions the built-in form field type
422 TYPE_REGEXP will recognize.
424 The configuration code prefers the POSIX regex facility, modeled on
425 System V's, but will settle for BSD regexps if the former is not
428 Historical note: the panels code was written primarily to assist in
429 porting u386mon 2.0 (comp.sources.misc v14i001-4) to systems lacking
430 panels support; u386mon 2.10 and beyond use it. This version has been
431 slightly cleaned up for ncurses.
433 A Tour of the Terminfo Compiler
435 The ncurses implementation of tic is rather complex internally; it has
436 to do a trying combination of missions. This starts with the fact
437 that, in addition to its normal duty of compiling terminfo sources
438 into loadable terminfo binaries, it has to be able to handle termcap
439 syntax and compile that too into terminfo entries.
441 The implementation therefore starts with a table-driven, dual-mode
442 lexical analyzer (in comp_scan.c). The lexer chooses its mode (termcap
443 or terminfo) based on the first "," or ":" it finds in each entry. The
444 lexer does all the work of recognizing capability names and values;
445 the grammar above it is trivial, just "parse entries till you run out
448 Translation of Non-use Capabilities
450 Translation of most things besides use capabilities is pretty
451 straightforward. The lexical analyzer's tokenizer hands each
452 capability name to a hash function, which drives a table lookup. The
453 table entry yields an index which is used to look up the token type in
454 another table, and controls interpretation of the value.
456 One possibly interesting aspect of the implementation is the way the
457 compiler tables are initialized. All the tables are generated by
458 various awk/sed/sh scripts from a master table include/Caps; these
459 scripts actually write C initializers which are linked to the
460 compiler. Furthermore, the hash table is generated in the same way, so
461 it doesn't have to be generated at compiler startup time (another
462 benefit of this organization is that the hash table can be in
463 shareable text space).
465 Thus, adding a new capability is usually pretty trivial, just a matter
466 of adding one line to the include/Caps file. We will have more to say
467 about this in the section on Source-Form Translation.
469 Use Capability Resolution
471 The background problem that makes tic tricky is not the capability
472 translation itself, it is the resolution of use capabilities. Older
473 versions would not handle forward use references for this reason (that
474 is, a using terminal always had to follow its use target in the source
475 file). By doing this, they got away with a simple implementation
476 tactic; compile everything as it blows by, then resolve uses from
479 This will not do for ncurses. The problem is that that the whole
480 compilation process has to be embeddable in the ncurses library so
481 that it can be called by the startup code to translate termcap entries
482 on the fly. The embedded version cannot go promiscuously writing
483 everything it translates out to disk -- for one thing, it will
484 typically be running with non-root permissions.
486 So our tic is designed to parse an entire terminfo file into a
487 doubly-linked circular list of entry structures in-core, and then do
488 use resolution in-memory before writing everything out. This design
489 has other advantages: it makes forward and back use-references equally
490 easy (so we get the latter for free), and it makes checking for name
491 collisions before they are written out easy to do.
493 And this is exactly how the embedded version works. But the
494 stand-alone user-accessible version of tic partly reverts to the
495 historical strategy; it writes to disk (not keeping in core) any entry
496 with no use references.
498 This is strictly a core-economy kluge, implemented because the
499 terminfo master file is large enough that some core-poor systems swap
500 like crazy when you compile it all in memory...there have been reports
501 of this process taking three hours, rather than the twenty seconds or
502 less typical on the author's development box.
504 So. The executable tic passes the entry-parser a hook that immediately
505 writes out the referenced entry if it has no use capabilities. The
506 compiler main loop refrains from adding the entry to the in-core list
507 when this hook fires. If some other entry later needs to reference an
508 entry that got written immediately, that is OK; the resolution code
509 will fetch it off disk when it cannot find it in core.
511 Name collisions will still be detected, just not as cleanly. The
512 write_entry() code complains before overwriting an entry that
513 postdates the time of tic's first call to write_entry(), Thus it will
514 complain about overwriting entries newly made during the tic run, but
515 not about overwriting ones that predate it.
517 Source-Form Translation
519 Another use of tic is to do source translation between various termcap
520 and terminfo formats. There are more variants out there than you might
521 think; the ones we know about are described in the captoinfo(1) manual
524 The translation output code (dump_entry() in ncurses/dump_entry.c) is
525 shared with the infocmp(1) utility. It takes the same internal
526 representation used to generate the binary form and dumps it to
527 standard output in a specified format.
529 The include/Caps file has a header comment describing ways you can
530 specify source translations for nonstandard capabilities just by
531 altering the master table. It is possible to set up capability
532 aliasing or tell the compiler to plain ignore a given capability
533 without writing any C code at all.
535 For circumstances where you need to do algorithmic translation, there
536 are functions in parse_entry.c called after the parse of each entry
537 that are specifically intended to encapsulate such translations. This,
538 for example, is where the AIX box1 capability get translated to an
543 The infocmp utility is just a wrapper around the same entry-dumping
544 code used by tic for source translation. Perhaps the one interesting
545 aspect of the code is the use of a predicate function passed in to
546 dump_entry() to control which capabilities are dumped. This is
547 necessary in order to handle both the ordinary De-compilation case and
548 entry difference reporting.
550 The tput and clear utilities just do an entry load followed by a
551 tputs() of a selected capability.
553 Style Tips for Developers
555 See the TO-DO file in the top-level directory of the source
556 distribution for additions that would be particularly useful.
558 The prefix _nc_ should be used on library public functions that are
559 not part of the curses API in order to prevent pollution of the
560 application namespace. If you have to add to or modify the function
561 prototypes in curses.h.in, read ncurses/MKlib_gen.sh first so you can
562 avoid breaking XSI conformance. Please join the ncurses mailing list.
563 See the INSTALL file in the top level of the distribution for details
566 Look for the string FIXME in source files to tag minor bugs and
567 potential problems that could use fixing.
569 Do not try to auto-detect OS features in the main body of the C code.
570 That is the job of the configuration system.
572 To hold down complexity, do make your code data-driven. Especially, if
573 you can drive logic from a table filtered out of include/Caps, do it.
574 If you find you need to augment the data in that file in order to
575 generate the proper table, that is still preferable to ad-hoc code --
576 that is why the fifth field (flags) is there.
582 The following notes are intended to be a first step towards DOS and
583 Macintosh ports of the ncurses libraries.
585 The following library modules are "pure curses"; they operate only on
586 the curses internal structures, do all output through other curses
587 calls (not including tputs() and putp()) and do not call any other
588 UNIX routines such as signal(2) or the stdio library. Thus, they
589 should not need to be modified for single-terminal ports.
591 lib_addch.c lib_addstr.c lib_bkgd.c lib_box.c lib_clear.c
592 lib_clrbot.c lib_clreol.c lib_delch.c lib_delwin.c lib_erase.c
593 lib_inchstr.c lib_insch.c lib_insdel.c lib_insstr.c lib_keyname.c
594 lib_move.c lib_mvwin.c lib_newwin.c lib_overlay.c lib_pad.c
595 lib_printw.c lib_refresh.c lib_scanw.c lib_scroll.c lib_scrreg.c
596 lib_set_term.c lib_touch.c lib_tparm.c lib_tputs.c lib_unctrl.c
599 This module is pure curses, but calls outstr():
603 These modules are pure curses, except that they use tputs() and
606 lib_beep.c lib_color.c lib_endwin.c lib_options.c lib_slk.c
609 This modules assist in POSIX emulation on non-POSIX systems:
614 The following source files will not be needed for a
615 single-terminal-type port.
617 alloc_entry.c captoinfo.c clear.c comp_captab.c comp_error.c
618 comp_hash.c comp_main.c comp_parse.c comp_scan.c dump_entry.c
619 infocmp.c parse_entry.c read_entry.c tput.c write_entry.c
621 The following modules will use open()/read()/write()/close()/lseek()
622 on files, but no other OS calls.
625 used to read/write screen dumps
628 used to write trace data to the logfile
630 Modules that would have to be modified for a port start here:
632 The following modules are "pure curses" but contain assumptions
633 inappropriate for a memory-mapped port.
636 assumes there may be multiple terminals
639 assumes acs_map as a double indirection
642 assumes cursor moves have variable cost
645 assumes there may be multiple terminals
648 assumes there may be multiple terminals
650 The following modules use UNIX-specific calls:
664 various tty-manipulation and system calls
667 various tty-manipulation calls
670 various tty-manipulation calls
673 various tty-manipulation calls
676 signal-manipulation calls
679 gettimeofday(), select().
680 _________________________________________________________________
683 Eric S. Raymond <esr@snark.thyrsus.com>
685 (Note: This is not the bug address!)