1 # Configuration variables of Sherlock and their default values
2 # (c) 2005 Martin Mares <mj@ucw.cz>
4 # Version of the whole package
5 Set("SHERLOCK_VERSION" => "3.5");
7 # Installation directory for `make install'
8 Set("INSTALL_DIR" => "~/run-3.5");
10 # Compile everything with debug information and ASSERT's
11 UnSet("CONFIG_DEBUG");
13 # Enable aggressive optimizations depending on exact CPU type (don't use for portable packages)
14 UnSet("CONFIG_EXACT_CPU");
16 ### Standard modules available in all releases of Sherlock ###
18 # Gatherer (the gatherer library and some utilities) [FIXME: decide automatically?]
19 Set("CONFIG_GATHERER");
21 # The gatherer daemon (requires CONFIG_GATHERER)
22 Set("CONFIG_GATHERD");
25 Set("CONFIG_INDEXER");
30 # Watson: The Monitoring System
31 UnSet("CONFIG_WATSON");
34 ### These are available in the commercial version only: ###
36 # The Shepherd daemon (requires CONFIG_GATHERER and CONFIG_SHEPHERD_PROTOCOL)
37 UnSet("CONFIG_SHEPHERD");
39 # Support for Shepherd protocols everywhere (distributed indexing etc.)
40 UnSet("CONFIG_SHEPHERD_PROTOCOL");
42 # The reaper daemon, used by Shepherd to download the pages (requires CONFIG_GATHERER)
43 UnSet("CONFIG_REAPD");
50 ### Optional Modules & Features ###
52 # Number of sections we remember limits for
53 Set("SHERLOCK_NUM_SECTIONS" => 4);
55 # Use shared libraries
56 UnSet("CONFIG_SHARED");
58 # Gathering of PDF files
61 # Searching by last modification time
62 Set("CONFIG_LASTMOD");
64 # Searching by file type
65 Set("CONFIG_FILETYPE");
67 # Language processing (requires CONFIG_FILETYPE)
73 # Enable indexing of context-dependent words and set maximum number
74 # of context slots (undef, 256 or 32768). Lower values save memory, higher time.
75 Set("CONFIG_MAX_CONTEXTS" => 256);
77 # Enable more than 16-bit word positions in references
78 Set("CONFIG_32BIT_REFERENCES");
81 UnSet("CONFIG_DEBUG_TOOLS");
83 # Search server: Enable EXPLAIN command (slows searching down slightly)
84 Set("CONFIG_EXPLAIN");
86 # If your system doesn't contain GNU libc 2.3 or newer, it's recommended to let Sherlock
87 # use its own regex library (a copy of the glibc one), because the default regex library
88 # is likely to be crappy.
89 Set("CONFIG_OWN_REGEX");
91 # Include custom builtin functions in filtering language
92 UnSet("CONFIG_CUSTOM_FILTER");
94 # Install libraries and their API includes
95 UnSet("CONFIG_INSTALL_API");
98 # Number of refresh frequencies we allow
99 Set("SHERLOCK_NUM_FREQS" => 5);
101 # Gathering and indexing split to areas
102 UnSet("CONFIG_AREAS");
105 UnSet("CONFIG_SITES");
107 # Gathering of images
108 UnSet("CONFIG_IMAGES");
110 # Gathering of MS Word files
111 UnSet("CONFIG_MSWORD");
113 # Gathering of MS Excel files
114 UnSet("CONFIG_EXCEL");
116 # Extra stuff for Centrum
117 UnSet("CONFIG_CENTRUM");
118 UnSet("CONFIG_CENTRUM_CZ");
119 UnSet("CONFIG_CENTRUM_SK");
120 UnSet("CONFIG_CENTRUM_PL");
122 # Debugging version of Centrum installation (lower limits, trivial filters etc.)
123 UnSet("CONFIG_CENTRUM_DEBUG");
125 # Stemmer by UFAL and its dictionaries
126 UnSet("CONFIG_UFAL_STEMMER");
127 UnSet("CONFIG_UFAL_DICT_CS");
128 UnSet("CONFIG_UFAL_DICT_SK");
130 # Stemmer for Polish with its dictionary
131 UnSet("CONFIG_PL_STEMMER");
134 UnSet("CONFIG_WEIGHTS");
138 ### These are internal settings which shouldn't be modified by custom configs ###
140 # Default configuration file
141 Set("DEFAULT_CONFIG" => "cf/sherlock");
143 # Configuration of parts of the UCW library
144 Set("CONFIG_UCW_PERL" => 1);
145 Set("CONFIG_UCW_SHELL_UTILS" => 1);
147 # Configuration of the language library
148 Set("CONFIG_LANG_DETECT" => 1);
150 # We always want to be able to use large files
151 Set("CONFIG_LFS" => 1);
153 # Maximum word length common for Sherlock and liblang, see comments in sherlock/index.h
154 Set("MAX_WORD_LEN" => 64);