2 $Id: hackguide.html,v 1.30 2017/05/05 11:50:09 tom Exp $
3 ****************************************************************************
4 * Copyright (c) 1998-2012,2017 Free Software Foundation, Inc. *
6 * Permission is hereby granted, free of charge, to any person obtaining a *
7 * copy of this software and associated documentation files (the *
8 * "Software"), to deal in the Software without restriction, including *
9 * without limitation the rights to use, copy, modify, merge, publish, *
10 * distribute, distribute with modifications, sublicense, and/or sell *
11 * copies of the Software, and to permit persons to whom the Software is *
12 * furnished to do so, subject to the following conditions: *
14 * The above copyright notice and this permission notice shall be included *
15 * in all copies or substantial portions of the Software. *
17 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS *
18 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF *
19 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. *
20 * IN NO EVENT SHALL THE ABOVE COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, *
21 * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR *
22 * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR *
23 * THE USE OR OTHER DEALINGS IN THE SOFTWARE. *
25 * Except as contained in this notice, the name(s) of the above copyright *
26 * holders shall not be used in advertising or otherwise to promote the *
27 * sale, use or other dealings in this Software without prior written *
29 ****************************************************************************
31 <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01//EN">
35 <meta name="generator" content=
36 "HTML Tidy for Linux (vers 25 March 2009), see www.w3.org">
38 <title>A Hacker's Guide to Ncurses Internals</title>
39 <link rev="made" href="mailto:bugs-ncurses@gnu.org">
40 <meta http-equiv="Content-Type" content=
41 "text/html; charset=us-ascii"><!--
42 This document is self-contained, *except* that there is one relative link to
43 the ncurses-intro.html document, expected to be in the same directory with
49 <h1>A Hacker's Guide to NCURSES</h1>
54 <li><a href="#abstract">Abstract</a></li>
57 <a href="#objective">Objective of the Package</a>
60 <li><a href="#whysvr4">Why System V Curses?</a></li>
62 <li><a href="#extensions">How to Design Extensions</a></li>
66 <li><a href="#portability">Portability and
67 Configuration</a></li>
69 <li><a href="#documentation">Documentation Conventions</a></li>
71 <li><a href="#bugtrack">How to Report Bugs</a></li>
74 <a href="#ncurslib">A Tour of the Ncurses Library</a>
77 <li><a href="#loverview">Library Overview</a></li>
79 <li><a href="#engine">The Engine Room</a></li>
81 <li><a href="#input">Keyboard Input</a></li>
83 <li><a href="#mouse">Mouse Events</a></li>
85 <li><a href="#output">Output and Screen Updating</a></li>
89 <li><a href="#fmnote">The Forms and Menu Libraries</a></li>
92 <a href="#tic">A Tour of the Terminfo Compiler</a>
95 <li><a href="#nonuse">Translation of
96 Non-<strong>use</strong> Capabilities</a></li>
98 <li><a href="#uses">Use Capability Resolution</a></li>
100 <li><a href="#translation">Source-Form Translation</a></li>
104 <li><a href="#utils">Other Utilities</a></li>
106 <li><a href="#style">Style Tips for Developers</a></li>
108 <li><a href="#port">Porting Hints</a></li>
111 <h1><a name="abstract" id="abstract">Abstract</a></h1>
113 <p>This document is a hacker's tour of the
114 <strong>ncurses</strong> library and utilities. It discusses
115 design philosophy, implementation methods, and the conventions
116 used for coding and documentation. It is recommended reading for
117 anyone who is interested in porting, extending or improving the
120 <h1><a name="objective" id="objective">Objective of the
123 <p>The objective of the <strong>ncurses</strong> package is to
124 provide a free software API for character-cell terminals and
125 terminal emulators with the following characteristics:</p>
128 <li>Source-compatible with historical curses implementations
129 (including the original BSD curses and System V curses.</li>
131 <li>Conformant with the XSI Curses standard issued as part of
134 <li>High-quality — stable and reliable code, wide
135 portability, good packaging, superior documentation.</li>
137 <li>Featureful — should eliminate as much of the drudgery
138 of C interface programming as possible, freeing programmers to
139 think at a higher level of design.</li>
142 <p>These objectives are in priority order. So, for example,
143 source compatibility with older version must trump featurefulness
144 — we cannot add features if it means breaking the portion
145 of the API corresponding to historical curses versions.</p>
147 <h2><a name="whysvr4" id="whysvr4">Why System V Curses?</a></h2>
149 <p>We used System V curses as a model, reverse-engineering their
150 API, in order to fulfill the first two objectives.</p>
152 <p>System V curses implementations can support BSD curses
153 programs with just a recompilation, so by capturing the System V
154 API we also capture BSD's.</p>
156 <p>More importantly for the future, the XSI Curses standard
157 issued by X/Open is explicitly and closely modeled on System V.
158 So conformance with System V took us most of the way to
159 base-level XSI conformance.</p>
161 <h2><a name="extensions" id="extensions">How to Design
164 <p>The third objective (standards conformance) requires that it
165 be easy to condition source code using <strong>ncurses</strong>
166 so that the absence of nonstandard extensions does not break the
169 <p>Accordingly, we have a policy of associating with each
170 nonstandard extension a feature macro, so that ncurses client
171 code can use this macro to condition in or out the code that
172 requires the <strong>ncurses</strong> extension.</p>
174 <p>For example, there is a macro
175 <code>NCURSES_MOUSE_VERSION</code> which XSI Curses does not
176 define, but which is defined in the <strong>ncurses</strong>
177 library header. You can use this to condition the calls to the
180 <h1><a name="portability" id="portability">Portability and
181 Configuration</a></h1>
183 <p>Code written for <strong>ncurses</strong> may assume an
184 ANSI-standard C compiler and POSIX-compatible OS interface. It
185 may also assume the presence of a System-V-compatible
186 <em>select(2)</em> call.</p>
188 <p>We encourage (but do not require) developers to make the code
189 friendly to less-capable UNIX environments wherever possible.</p>
191 <p>We encourage developers to support OS-specific optimizations
192 and methods not available under POSIX/ANSI, provided only
196 <li>All such code is properly conditioned so the build process
197 does not attempt to compile it under a plain ANSI/POSIX
200 <li>Adding such implementation methods does not introduce
201 incompatibilities in the <strong>ncurses</strong> API between
205 <p>We use GNU <code>autoconf(1)</code> as a tool to deal with
206 portability issues. The right way to leverage an OS-specific
207 feature is to modify the autoconf specification files
208 (configure.in and aclocal.m4) to set up a new feature macro,
209 which you then use to condition your code.</p>
211 <h1><a name="documentation" id="documentation">Documentation
214 <p>There are three kinds of documentation associated with this
215 package. Each has a different preferred format:</p>
218 <li>Package-internal files (README, INSTALL, TO-DO etc.)</li>
220 <li>Manual pages.</li>
222 <li>Everything else (i.e., narrative documentation).</li>
225 <p>Our conventions are simple:</p>
228 <li><strong>Maintain package-internal files in plain
229 text.</strong> The expected viewer for them <em>more(1)</em> or
230 an editor window; there is no point in elaborate mark-up.</li>
232 <li><strong>Mark up manual pages in the man macros.</strong>
233 These have to be viewable through traditional <em>man(1)</em>
236 <li><strong>Write everything else in HTML.</strong></li>
239 <p>When in doubt, HTMLize a master and use <em>lynx(1)</em> to
240 generate plain ASCII (as we do for the announcement
243 <p>The reason for choosing HTML is that it is (a) well-adapted
244 for on-line browsing through viewers that are everywhere; (b)
245 more easily readable as plain text than most other mark-ups, if
246 you do not have a viewer; and (c) carries enough information that
247 you can generate a nice-looking printed version from it. Also, of
248 course, it make exporting things like the announcement document
249 to WWW pretty trivial.</p>
251 <h1><a name="bugtrack" id="bugtrack">How to Report Bugs</a></h1>
253 <p>The <a name="bugreport" id="bugreport">reporting address for
255 "mailto:bug-ncurses@gnu.org">bug-ncurses@gnu.org</a>. This is a
256 majordomo list; to join, write to
257 <code>bug-ncurses-request@gnu.org</code> with a message
258 containing the line:</p>
260 subscribe <name>@<host.domain>
263 <p>The <code>ncurses</code> code is maintained by a small group
264 of volunteers. While we try our best to fix bugs promptly, we
265 simply do not have a lot of hours to spend on elementary
266 hand-holding. We rely on intelligent cooperation from our users.
267 If you think you have found a bug in <code>ncurses</code>, there
268 are some steps you can take before contacting us that will help
269 get the bug fixed quickly.</p>
271 <p>In order to use our bug-fixing time efficiently, we put people
272 who show us they have taken these steps at the head of our queue.
273 This means that if you do not, you will probably end up at the
274 tail end and have to wait a while.</p>
277 <li>Develop a recipe to reproduce the bug.
279 <p>Bugs we can reproduce are likely to be fixed very quickly,
280 often within days. The most effective single thing you can do
281 to get a quick fix is develop a way we can duplicate the bad
282 behavior — ideally, by giving us source for a small,
283 portable test program that breaks the library. (Even better
284 is a keystroke recipe using one of the test programs provided
285 with the distribution.)</p>
288 <li>Try to reproduce the bug on a different terminal type.
290 <p>In our experience, most of the behaviors people report as
291 library bugs are actually due to subtle problems in terminal
292 descriptions. This is especially likely to be true if you are
293 using a traditional asynchronous terminal or PC-based
294 terminal emulator, rather than xterm or a UNIX console
297 <p>It is therefore extremely helpful if you can tell us
298 whether or not your problem reproduces on other terminal
299 types. Usually you will have both a console type and xterm
300 available; please tell us whether or not your bug reproduces
303 <p>If you have xterm available, it is also good to collect
304 xterm reports for different window sizes. This is especially
305 true if you normally use an unusual xterm window size —
306 a surprising number of the bugs we have seen are either
307 triggered or masked by these.</p>
310 <li>Generate and examine a trace file for the broken behavior.
312 <p>Recompile your program with the debugging versions of the
313 libraries. Insert a <code>trace()</code> call with the
314 argument set to <code>TRACE_UPDATE</code>. (See <a href=
315 "ncurses-intro.html#debugging">"Writing Programs with
316 NCURSES"</a> for details on trace levels.) Reproduce your
317 bug, then look at the trace file to see what the library was
320 <p>Another frequent cause of apparent bugs is application
321 coding errors that cause the wrong things to be put on the
322 virtual screen. Looking at the virtual-screen dumps in the
323 trace file will tell you immediately if this is happening,
324 and save you from the possible embarrassment of being told
325 that the bug is in your code and is your problem rather than
328 <p>If the virtual-screen dumps look correct but the bug
329 persists, it is possible to crank up the trace level to give
330 more and more information about the library's update actions
331 and the control sequences it issues to perform them. The test
332 directory of the distribution contains a tool for digesting
333 these logs to make them less tedious to wade through.</p>
335 <p>Often you will find terminfo problems at this stage by
336 noticing that the escape sequences put out for various
337 capabilities are wrong. If not, you are likely to learn
338 enough to be able to characterize any bug in the
339 screen-update logic quite exactly.</p>
342 <li>Report details and symptoms, not just interpretations.
344 <p>If you do the preceding two steps, it is very likely that
345 you will discover the nature of the problem yourself and be
346 able to send us a fix. This will create happy feelings all
347 around and earn you good karma for the first time you run
348 into a bug you really cannot characterize and fix
351 <p>If you are still stuck, at least you will know what to
352 tell us. Remember, we need details. If you guess about what
353 is safe to leave out, you are too likely to be wrong.</p>
355 <p>If your bug produces a bad update, include a trace file.
356 Try to make the trace at the <em>least</em> voluminous level
357 that pins down the bug. Logs that have been through
358 tracemunch are OK, it does not throw away any information
359 (actually they are better than un-munched ones because they
360 are easier to read).</p>
362 <p>If your bug produces a core-dump, please include a
363 symbolic stack trace generated by gdb(1) or your local
366 <p>Tell us about every terminal on which you have reproduced
367 the bug — and every terminal on which you cannot.
368 Ideally, sent us terminfo sources for all of these (yours
369 might differ from ours).</p>
371 <p>Include your ncurses version and your OS/machine type, of
372 course! You can find your ncurses version in the
373 <code>curses.h</code> file.</p>
377 <p>If your problem smells like a logic error or in cursor
378 movement or scrolling or a bad capability, there are a couple of
379 tiny test frames for the library algorithms in the progs
380 directory that may help you isolate it. These are not part of the
381 normal build, but do have their own make productions.</p>
383 <p>The most important of these is <code>mvcur</code>, a test
384 frame for the cursor-movement optimization code. With this
385 program, you can see directly what control sequences will be
386 emitted for any given cursor movement or scroll/insert/delete
387 operations. If you think you have got a bad capability
388 identified, you can disable it and test again. The program is
389 command-driven and has on-line help.</p>
391 <p>If you think the vertical-scroll optimization is broken, or
392 just want to understand how it works better, build
393 <code>hashmap</code> and read the header comments of
394 <code>hardscroll.c</code> and <code>hashmap.c</code>; then try it
395 out. You can also test the hardware-scrolling optimization
396 separately with <code>hardscroll</code>.</p>
398 <h1><a name="ncurslib" id="ncurslib">A Tour of the Ncurses
401 <h2><a name="loverview" id="loverview">Library Overview</a></h2>
403 <p>Most of the library is superstructure — fairly trivial
404 convenience interfaces to a small set of basic functions and data
405 structures used to manipulate the virtual screen (in particular,
406 none of this code does any I/O except through calls to more
407 fundamental modules described below). The files</p>
410 <code>lib_addch.c lib_bkgd.c lib_box.c lib_chgat.c lib_clear.c
411 lib_clearok.c lib_clrbot.c lib_clreol.c lib_colorset.c
412 lib_data.c lib_delch.c lib_delwin.c lib_echo.c lib_erase.c
413 lib_gen.c lib_getstr.c lib_hline.c lib_immedok.c lib_inchstr.c
414 lib_insch.c lib_insdel.c lib_insstr.c lib_instr.c
415 lib_isendwin.c lib_keyname.c lib_leaveok.c lib_move.c
416 lib_mvwin.c lib_overlay.c lib_pad.c lib_printw.c lib_redrawln.c
417 lib_scanw.c lib_screen.c lib_scroll.c lib_scrollok.c
418 lib_scrreg.c lib_set_term.c lib_slk.c lib_slkatr_set.c
419 lib_slkatrof.c lib_slkatron.c lib_slkatrset.c lib_slkattr.c
420 lib_slkclear.c lib_slkcolor.c lib_slkinit.c lib_slklab.c
421 lib_slkrefr.c lib_slkset.c lib_slktouch.c lib_touch.c
422 lib_unctrl.c lib_vline.c lib_wattroff.c lib_wattron.c
426 <p>are all in this category. They are very unlikely to need
427 change, barring bugs or some fundamental reorganization in the
428 underlying data structures.</p>
430 <p>These files are used only for debugging support:</p>
433 <code>lib_trace.c lib_traceatr.c lib_tracebits.c lib_tracechr.c
434 lib_tracedmp.c lib_tracemse.c trace_buf.c</code>
437 <p>It is rather unlikely you will ever need to change these,
438 unless you want to introduce a new debug trace level for some
441 <p>There is another group of files that do direct I/O via
442 <em>tputs()</em>, computations on the terminal capabilities, or
443 queries to the OS environment, but nevertheless have only fairly
444 low complexity. These include:</p>
447 <code>lib_acs.c lib_beep.c lib_color.c lib_endwin.c
448 lib_initscr.c lib_longname.c lib_newterm.c lib_options.c
449 lib_termcap.c lib_ti.c lib_tparm.c lib_tputs.c lib_vidattr.c
453 <p>They are likely to need revision only if ncurses is being
454 ported to an environment without an underlying terminfo
455 capability representation.</p>
457 <p>These files have serious hooks into the tty driver and signal
461 <code>lib_kernel.c lib_baudrate.c lib_raw.c lib_tstp.c
465 <p>If you run into porting snafus moving the package to another
466 UNIX, the problem is likely to be in one of these files. The file
467 <code>lib_print.c</code> uses sleep(2) and also falls in this
470 <p>Almost all of the real work is done in the files</p>
473 <code>hardscroll.c hashmap.c lib_addch.c lib_doupdate.c
474 lib_getch.c lib_mouse.c lib_mvcur.c lib_refresh.c lib_setup.c
478 <p>Most of the algorithmic complexity in the library lives in
479 these files. If there is a real bug in <strong>ncurses</strong>
480 itself, it is probably here. We will tour some of these files in
481 detail below (see <a href="#engine">The Engine Room</a>).</p>
483 <p>Finally, there is a group of files that is actually most of
484 the terminfo compiler. The reason this code lives in the
485 <strong>ncurses</strong> library is to support fallback to
486 /etc/termcap. These files include</p>
489 <code>alloc_entry.c captoinfo.c comp_captab.c comp_error.c
490 comp_hash.c comp_parse.c comp_scan.c parse_entry.c
491 read_termcap.c write_entry.c</code>
494 <p>We will discuss these in the compiler tour.</p>
496 <h2><a name="engine" id="engine">The Engine Room</a></h2>
498 <h3><a name="input" id="input">Keyboard Input</a></h3>
500 <p>All <code>ncurses</code> input funnels through the function
501 <code>wgetch()</code>, defined in <code>lib_getch.c</code>. This
502 function is tricky; it has to poll for keyboard and mouse events
503 and do a running match of incoming input against the set of
504 defined special keys.</p>
506 <p>The central data structure in this module is a FIFO queue,
507 used to match multiple-character input sequences against
508 special-key capabilities; also to implement pushback via
509 <code>ungetch()</code>.</p>
511 <p>The <code>wgetch()</code> code distinguishes between function
512 key sequences and the same sequences typed manually by doing a
513 timed wait after each input character that could lead a function
514 key sequence. If the entire sequence takes less than 1 second, it
515 is assumed to have been generated by a function key press.</p>
517 <p>Hackers bruised by previous encounters with variant
518 <code>select(2)</code> calls may find the code in
519 <code>lib_twait.c</code> interesting. It deals with the problem
520 that some BSD selects do not return a reliable time-left value.
521 The function <code>timed_wait()</code> effectively simulates a
524 <h3><a name="mouse" id="mouse">Mouse Events</a></h3>
526 <p>If the mouse interface is active, <code>wgetch()</code> polls
527 for mouse events each call, before it goes to the keyboard for
528 input. It is up to <code>lib_mouse.c</code> how the polling is
529 accomplished; it may vary for different devices.</p>
531 <p>Under xterm, however, mouse event notifications come in via
532 the keyboard input stream. They are recognized by having the
533 <strong>kmous</strong> capability as a prefix. This is kind of
534 klugey, but trying to wire in recognition of a mouse key prefix
535 without going through the function-key machinery would be just
536 too painful, and this turns out to imply having the prefix
537 somewhere in the function-key capabilities at terminal-type
540 <p>This kluge only works because <strong>kmous</strong> is not
541 actually used by any historic terminal type or curses
542 implementation we know of. Best guess is it is a relic of some
543 forgotten experiment in-house at Bell Labs that did not leave any
544 traces in the publicly-distributed System V terminfo files. If
545 System V or XPG4 ever gets serious about using it again, this
546 kluge may have to change.</p>
548 <p>Here are some more details about mouse event handling:</p>
550 <p>The <code>lib_mouse()</code>code is logically split into a
551 lower level that accepts event reports in a device-dependent
552 format and an upper level that parses mouse gestures and filters
553 events. The mediating data structure is a circular queue of event
556 <p>Functionally, the lower level's job is to pick up primitive
557 events and put them on the circular queue. This can happen in one
558 of two ways: either (a) <code>_nc_mouse_event()</code> detects a
559 series of incoming mouse reports and queues them, or (b) code in
560 <code>lib_getch.c</code> detects the <strong>kmous</strong>
561 prefix in the keyboard input stream and calls _nc_mouse_inline to
562 queue up a series of adjacent mouse reports.</p>
564 <p>In either case, <code>_nc_mouse_parse()</code> should be
565 called after the series is accepted to parse the digested mouse
566 reports (low-level events) into a gesture (a high-level or
567 composite event).</p>
569 <h3><a name="output" id="output">Output and Screen
572 <p>With the single exception of character echoes during a
573 <code>wgetnstr()</code> call (which simulates cooked-mode line
574 editing in an ncurses window), the library normally does all its
575 output at refresh time.</p>
577 <p>The main job is to go from the current state of the screen (as
578 represented in the <code>curscr</code> window structure) to the
579 desired new state (as represented in the <code>newscr</code>
580 window structure), while doing as little I/O as possible.</p>
582 <p>The brains of this operation are the modules
583 <code>hashmap.c</code>, <code>hardscroll.c</code> and
584 <code>lib_doupdate.c</code>; the latter two use
585 <code>lib_mvcur.c</code>. Essentially, what happens looks like
588 <p>The <code>hashmap.c</code> module tries to detect vertical
589 motion changes between the real and virtual screens. This
590 information is represented by the oldindex members in the newscr
591 structure. These are modified by vertical-motion and clear
592 operations, and both are re-initialized after each update. To
593 this change-journalling information, the hashmap code adds
594 deductions made using a modified Heckel algorithm on hash values
595 generated from the line contents.</p>
597 <p>The <code>hardscroll.c</code> module computes an optimum set
598 of scroll, insertion, and deletion operations to make the indices
599 match. It calls <code>_nc_mvcur_scrolln()</code> in
600 <code>lib_mvcur.c</code> to do those motions.</p>
602 <p>Then <code>lib_doupdate.c</code> goes to work. Its job is to
603 do line-by-line transformations of <code>curscr</code> lines to
604 <code>newscr</code> lines. Its main tool is the routine
605 <code>mvcur()</code> in <code>lib_mvcur.c</code>. This routine
606 does cursor-movement optimization, attempting to get from given
607 screen location A to given location B in the fewest output
608 characters possible.</p>
610 <p>If you want to work on screen optimizations, you should use
611 the fact that (in the trace-enabled version of the library)
612 enabling the <code>TRACE_TIMES</code> trace level causes a report
613 to be emitted after each screen update giving the elapsed time
614 and a count of characters emitted during the update. You can use
615 this to tell when an update optimization improves efficiency.</p>
617 <p>In the trace-enabled version of the library, it is also
618 possible to disable and re-enable various optimizations at
619 runtime by tweaking the variable
620 <code>_nc_optimize_enable</code>. See the file
621 <code>include/curses.h.in</code> for mask values, near the
624 <h1><a name="fmnote" id="fmnote">The Forms and Menu
627 <p>The forms and menu libraries should work reliably in any
628 environment you can port ncurses to. The only portability issue
629 anywhere in them is what flavor of regular expressions the
630 built-in form field type TYPE_REGEXP will recognize.</p>
632 <p>The configuration code prefers the POSIX regex facility,
633 modeled on System V's, but will settle for BSD regexps if the
634 former is not available.</p>
636 <p>Historical note: the panels code was written primarily to
637 assist in porting u386mon 2.0 (comp.sources.misc v14i001-4) to
638 systems lacking panels support; u386mon 2.10 and beyond use it.
639 This version has been slightly cleaned up for
640 <code>ncurses</code>.</p>
642 <h1><a name="tic" id="tic">A Tour of the Terminfo
645 <p>The <strong>ncurses</strong> implementation of
646 <strong>tic</strong> is rather complex internally; it has to do a
647 trying combination of missions. This starts with the fact that,
648 in addition to its normal duty of compiling terminfo sources into
649 loadable terminfo binaries, it has to be able to handle termcap
650 syntax and compile that too into terminfo entries.</p>
652 <p>The implementation therefore starts with a table-driven,
653 dual-mode lexical analyzer (in <code>comp_scan.c</code>). The
654 lexer chooses its mode (termcap or terminfo) based on the first
655 “,” or “:” it finds in each entry. The
656 lexer does all the work of recognizing capability names and
657 values; the grammar above it is trivial, just "parse entries till
658 you run out of file".</p>
660 <h2><a name="nonuse" id="nonuse">Translation of
661 Non-<strong>use</strong> Capabilities</a></h2>
663 <p>Translation of most things besides <strong>use</strong>
664 capabilities is pretty straightforward. The lexical analyzer's
665 tokenizer hands each capability name to a hash function, which
666 drives a table lookup. The table entry yields an index which is
667 used to look up the token type in another table, and controls
668 interpretation of the value.</p>
670 <p>One possibly interesting aspect of the implementation is the
671 way the compiler tables are initialized. All the tables are
672 generated by various awk/sed/sh scripts from a master table
673 <code>include/Caps</code>; these scripts actually write C
674 initializers which are linked to the compiler. Furthermore, the
675 hash table is generated in the same way, so it doesn't have to be
676 generated at compiler startup time (another benefit of this
677 organization is that the hash table can be in shareable text
680 <p>Thus, adding a new capability is usually pretty trivial, just
681 a matter of adding one line to the <code>include/Caps</code>
682 file. We will have more to say about this in the section on
683 <a href="#translation">Source-Form Translation</a>.</p>
685 <h2><a name="uses" id="uses">Use Capability Resolution</a></h2>
687 <p>The background problem that makes <strong>tic</strong> tricky
688 is not the capability translation itself, it is the resolution of
689 <strong>use</strong> capabilities. Older versions would not
690 handle forward <strong>use</strong> references for this reason
691 (that is, a using terminal always had to follow its use target in
692 the source file). By doing this, they got away with a simple
693 implementation tactic; compile everything as it blows by, then
694 resolve uses from compiled entries.</p>
696 <p>This will not do for <strong>ncurses</strong>. The problem is
697 that that the whole compilation process has to be embeddable in
698 the <strong>ncurses</strong> library so that it can be called by
699 the startup code to translate termcap entries on the fly. The
700 embedded version cannot go promiscuously writing everything it
701 translates out to disk — for one thing, it will typically
702 be running with non-root permissions.</p>
704 <p>So our <strong>tic</strong> is designed to parse an entire
705 terminfo file into a doubly-linked circular list of entry
706 structures in-core, and then do <strong>use</strong> resolution
707 in-memory before writing everything out. This design has other
708 advantages: it makes forward and back use-references equally easy
709 (so we get the latter for free), and it makes checking for name
710 collisions before they are written out easy to do.</p>
712 <p>And this is exactly how the embedded version works. But the
713 stand-alone user-accessible version of <strong>tic</strong>
714 partly reverts to the historical strategy; it writes to disk (not
715 keeping in core) any entry with no <strong>use</strong>
718 <p>This is strictly a core-economy kluge, implemented because the
719 terminfo master file is large enough that some core-poor systems
720 swap like crazy when you compile it all in memory...there have
721 been reports of this process taking <strong>three hours</strong>,
722 rather than the twenty seconds or less typical on the author's
725 <p>So. The executable <strong>tic</strong> passes the
726 entry-parser a hook that <em>immediately</em> writes out the
727 referenced entry if it has no use capabilities. The compiler main
728 loop refrains from adding the entry to the in-core list when this
729 hook fires. If some other entry later needs to reference an entry
730 that got written immediately, that is OK; the resolution code
731 will fetch it off disk when it cannot find it in core.</p>
733 <p>Name collisions will still be detected, just not as cleanly.
734 The <code>write_entry()</code> code complains before overwriting
735 an entry that postdates the time of <strong>tic</strong>'s first
736 call to <code>write_entry()</code>, Thus it will complain about
737 overwriting entries newly made during the <strong>tic</strong>
738 run, but not about overwriting ones that predate it.</p>
740 <h2><a name="translation" id="translation">Source-Form
743 <p>Another use of <strong>tic</strong> is to do source
744 translation between various termcap and terminfo formats. There
745 are more variants out there than you might think; the ones we
746 know about are described in the <strong>captoinfo(1)</strong>
749 <p>The translation output code (<code>dump_entry()</code> in
750 <code>ncurses/dump_entry.c</code>) is shared with the
751 <strong>infocmp(1)</strong> utility. It takes the same internal
752 representation used to generate the binary form and dumps it to
753 standard output in a specified format.</p>
755 <p>The <code>include/Caps</code> file has a header comment
756 describing ways you can specify source translations for
757 nonstandard capabilities just by altering the master table. It is
758 possible to set up capability aliasing or tell the compiler to
759 plain ignore a given capability without writing any C code at
762 <p>For circumstances where you need to do algorithmic
763 translation, there are functions in <code>parse_entry.c</code>
764 called after the parse of each entry that are specifically
765 intended to encapsulate such translations. This, for example, is
766 where the AIX <strong>box1</strong> capability get translated to
767 an <strong>acsc</strong> string.</p>
769 <h1><a name="utils" id="utils">Other Utilities</a></h1>
771 <p>The <strong>infocmp</strong> utility is just a wrapper around
772 the same entry-dumping code used by <strong>tic</strong> for
773 source translation. Perhaps the one interesting aspect of the
774 code is the use of a predicate function passed in to
775 <code>dump_entry()</code> to control which capabilities are
776 dumped. This is necessary in order to handle both the ordinary
777 De-compilation case and entry difference reporting.</p>
779 <p>The <strong>tput</strong> and <strong>clear</strong> utilities
780 just do an entry load followed by a <code>tputs()</code> of a
781 selected capability.</p>
783 <h1><a name="style" id="style">Style Tips for Developers</a></h1>
785 <p>See the TO-DO file in the top-level directory of the source
786 distribution for additions that would be particularly useful.</p>
788 <p>The prefix <code>_nc_</code> should be used on library public
789 functions that are not part of the curses API in order to prevent
790 pollution of the application namespace. If you have to add to or
791 modify the function prototypes in curses.h.in, read
792 ncurses/MKlib_gen.sh first so you can avoid breaking XSI
793 conformance. Please join the ncurses mailing list. See the
794 INSTALL file in the top level of the distribution for details on
797 <p>Look for the string <code>FIXME</code> in source files to tag
798 minor bugs and potential problems that could use fixing.</p>
800 <p>Do not try to auto-detect OS features in the main body of the
801 C code. That is the job of the configuration system.</p>
803 <p>To hold down complexity, do make your code data-driven.
804 Especially, if you can drive logic from a table filtered out of
805 <code>include/Caps</code>, do it. If you find you need to augment
806 the data in that file in order to generate the proper table, that
807 is still preferable to ad-hoc code — that is why the fifth
808 field (flags) is there.</p>
812 <h1><a name="port" id="port">Porting Hints</a></h1>
814 <p>The following notes are intended to be a first step towards
815 DOS and Macintosh ports of the ncurses libraries.</p>
817 <p>The following library modules are “pure curses”;
818 they operate only on the curses internal structures, do all
819 output through other curses calls (not including
820 <code>tputs()</code> and <code>putp()</code>) and do not call any
821 other UNIX routines such as signal(2) or the stdio library. Thus,
822 they should not need to be modified for single-terminal
826 <code>lib_addch.c lib_addstr.c lib_bkgd.c lib_box.c lib_clear.c
827 lib_clrbot.c lib_clreol.c lib_delch.c lib_delwin.c lib_erase.c
828 lib_inchstr.c lib_insch.c lib_insdel.c lib_insstr.c
829 lib_keyname.c lib_move.c lib_mvwin.c lib_newwin.c lib_overlay.c
830 lib_pad.c lib_printw.c lib_refresh.c lib_scanw.c lib_scroll.c
831 lib_scrreg.c lib_set_term.c lib_touch.c lib_tparm.c lib_tputs.c
832 lib_unctrl.c lib_window.c panel.c</code>
835 <p>This module is pure curses, but calls outstr():</p>
838 <code>lib_getstr.c</code>
841 <p>These modules are pure curses, except that they use
842 <code>tputs()</code> and <code>putp()</code>:</p>
845 <code>lib_beep.c lib_color.c lib_endwin.c lib_options.c
846 lib_slk.c lib_vidattr.c</code>
849 <p>This modules assist in POSIX emulation on non-POSIX
855 <dd>signal calls</dd>
858 <p>The following source files will not be needed for a
859 single-terminal-type port.</p>
862 <code>alloc_entry.c captoinfo.c clear.c comp_captab.c
863 comp_error.c comp_hash.c comp_main.c comp_parse.c comp_scan.c
864 dump_entry.c infocmp.c parse_entry.c read_entry.c tput.c
868 <p>The following modules will use
869 open()/read()/write()/close()/lseek() on files, but no other OS
873 <dt>lib_screen.c</dt>
875 <dd>used to read/write screen dumps</dd>
879 <dd>used to write trace data to the logfile</dd>
882 <p>Modules that would have to be modified for a port start
885 <p>The following modules are “pure curses” but
886 contain assumptions inappropriate for a memory-mapped port.</p>
889 <dt>lib_longname.c</dt>
891 <dd>assumes there may be multiple terminals</dd>
895 <dd>assumes acs_map as a double indirection</dd>
899 <dd>assumes cursor moves have variable cost</dd>
901 <dt>lib_termcap.c</dt>
903 <dd>assumes there may be multiple terminals</dd>
907 <dd>assumes there may be multiple terminals</dd>
910 <p>The following modules use UNIX-specific calls:</p>
913 <dt>lib_doupdate.c</dt>
915 <dd>input checking</dd>
921 <dt>lib_initscr.c</dt>
925 <dt>lib_newterm.c</dt>
927 <dt>lib_baudrate.c</dt>
929 <dt>lib_kernel.c</dt>
931 <dd>various tty-manipulation and system calls</dd>
935 <dd>various tty-manipulation calls</dd>
939 <dd>various tty-manipulation calls</dd>
941 <dt>lib_restart.c</dt>
943 <dd>various tty-manipulation calls</dd>
947 <dd>signal-manipulation calls</dd>
951 <dd>gettimeofday(), select().</dd>
956 Eric S. Raymond <esr@snark.thyrsus.com>
957 </address>(Note: This is <em>not</em> the <a href="#bugtrack">bug