(url-open-stream): Really use asynchronous connections (accidentally
[emacs.git] / src / region-cache.h
blob40e68084cce2d87398e0a5676df6a19559608123
1 /* Header file: Caching facts about regions of the buffer, for optimization.
2 Copyright (C) 1985, 1986, 1993, 1995, 2002, 2003, 2004,
3 2005, 2006 Free Software Foundation, Inc.
5 This file is part of GNU Emacs.
7 GNU Emacs is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 2, or (at your option)
10 any later version.
12 GNU Emacs is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with GNU Emacs; see the file COPYING. If not, write to
19 the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
20 Boston, MA 02110-1301, USA. */
23 /* This code was written by Jim Blandy <jimb@cs.oberlin.edu> to help
24 GNU Emacs better support the gene editor written for the University
25 of Illinois at Urbana-Champagne's Ribosome Database Project (RDP).
27 Emacs implements line operations (finding the beginning/end of the
28 line, vertical motion, all the redisplay stuff) by searching for
29 newlines in the buffer. Usually, this is a good design; it's very
30 clean to just represent the buffer as an unstructured string of
31 characters, and the lines in most files are very short (less than
32 eighty characters), meaning that scanning usually costs about the
33 same as the overhead of maintaining some more complicated data
34 structure.
36 However, some applications, like gene editing, make use of very
37 long lines --- on the order of tens of kilobytes. In such cases,
38 it may well be worthwhile to try to avoid scanning, because the
39 scans have become two orders of magnitude more expensive. It would
40 be nice if this speedup could preserve the simplicity of the
41 existing data structure, and disturb as little of the existing code
42 as possible.
44 So here's the tack. We add some caching to the scan_buffer
45 function, so that when it searches for a newline, it notes that the
46 region between the start and end of the search contained no
47 newlines; then, the next time around, it consults this cache to see
48 if there are regions of text it can skip over completely. The
49 buffer modification primitives invalidate this cache.
51 (Note: Since the redisplay code needs similar information on
52 modified regions of the buffer, we can use the code that helps out
53 redisplay as a guide to where we need to add our own code to
54 invalidate our cache. prepare_to_modify_buffer seems to be the
55 central spot.)
57 Note that the cache code itself never mentions newlines
58 specifically, so if you wanted to cache other properties of regions
59 of the buffer, you could use this code pretty much unchanged. So
60 this cache really holds "known/unknown" information --- "I know
61 this region has property P" vs. "I don't know if this region has
62 property P or not." */
65 /* Allocate, initialize and return a new, empty region cache. */
66 struct region_cache *new_region_cache P_ ((void));
68 /* Free a region cache. */
69 void free_region_cache P_ ((struct region_cache *));
71 /* Assert that the region of BUF between START and END (absolute
72 buffer positions) is "known," for the purposes of CACHE (e.g. "has
73 no newlines", in the case of the line cache). */
74 extern void know_region_cache P_ ((struct buffer *BUF,
75 struct region_cache *CACHE,
76 int START, int END));
78 /* Indicate that a section of BUF has changed, to invalidate CACHE.
79 HEAD is the number of chars unchanged at the beginning of the buffer.
80 TAIL is the number of chars unchanged at the end of the buffer.
81 NOTE: this is *not* the same as the ending position of modified
82 region.
83 (This way of specifying regions makes more sense than absolute
84 buffer positions in the presence of insertions and deletions; the
85 args to pass are the same before and after such an operation.) */
86 extern void invalidate_region_cache P_ ((struct buffer *BUF,
87 struct region_cache *CACHE,
88 int HEAD, int TAIL));
90 /* The scanning functions.
92 Basically, if you're scanning forward/backward from position POS,
93 and region_cache_forward/backward returns true, you can skip all
94 the text between POS and *NEXT. And if the function returns false,
95 you should examine all the text from POS to *NEXT, and call
96 know_region_cache depending on what you find there; this way, you
97 might be able to avoid scanning it again. */
99 /* Return true if the text immediately after POS in BUF is known, for
100 the purposes of CACHE. If NEXT is non-zero, set *NEXT to the nearest
101 position after POS where the knownness changes. */
102 extern int region_cache_forward P_ ((struct buffer *BUF,
103 struct region_cache *CACHE,
104 int POS,
105 int *NEXT));
107 /* Return true if the text immediately before POS in BUF is known, for
108 the purposes of CACHE. If NEXT is non-zero, set *NEXT to the nearest
109 position before POS where the knownness changes. */
110 extern int region_cache_backward P_ ((struct buffer *BUF,
111 struct region_cache *CACHE,
112 int POS,
113 int *NEXT));
115 /* arch-tag: 70f79125-ef22-4f58-9aec-a48ca2791435
116 (do not change this comment) */