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 ### Standard modules available in all releases of Sherlock ###
15 # Gatherer (the gatherer library and some utilities) [FIXME: decide automatically?]
16 Set("CONFIG_GATHERER");
18 # The gatherer daemon (requires CONFIG_GATHERER)
19 Set("CONFIG_GATHERD");
22 Set("CONFIG_INDEXER");
27 # Watson: The Monitoring System
28 UnSet("CONFIG_WATSON");
31 ### These are available in the commercial version only: ###
33 # The Shepherd daemon (requires CONFIG_GATHERER and CONFIG_SHEPHERD_PROTOCOL)
34 UnSet("CONFIG_SHEPHERD");
36 # Support for Shepherd protocols everywhere (distributed indexing etc.)
37 UnSet("CONFIG_SHEPHERD_PROTOCOL");
39 # The reaper daemon, used by Shepherd to download the pages (requires CONFIG_GATHERER)
40 UnSet("CONFIG_REAPD");
47 ### Optional Modules & Features ###
49 # Number of sections we remember limits for
50 Set("SHERLOCK_NUM_SECTIONS" => 4);
52 # Use shared libraries
53 UnSet("CONFIG_SHARED");
55 # Gathering of PDF files
58 # Searching by last modification time
59 Set("CONFIG_LASTMOD");
61 # Searching by file type
62 Set("CONFIG_FILETYPE");
64 # Language processing (requires CONFIG_FILETYPE)
70 # Enable indexing of context-dependent words and set maximum number
71 # of context slots (undef, 256 or 32768). Lower values save memory, higher time.
72 Set("CONFIG_MAX_CONTEXTS" => 256);
74 # Enable more than 16-bit word positions in references
75 Set("CONFIG_32BIT_REFERENCES");
78 UnSet("CONFIG_DEBUG_TOOLS");
80 # Search server: Enable EXPLAIN command (slows searching down slightly)
81 Set("CONFIG_EXPLAIN");
83 # If your system doesn't contain GNU libc 2.3 or newer, it's recommended to let Sherlock
84 # use its own regex library (a copy of the glibc one), because the default regex library
85 # is likely to be crappy.
86 Set("CONFIG_OWN_REGEX");
88 # Include custom builtin functions in filtering language
89 UnSet("CONFIG_CUSTOM_FILTER");
91 # Install libraries and their API includes
92 UnSet("CONFIG_INSTALL_API");
95 # Number of refresh frequencies we allow
96 Set("SHERLOCK_NUM_FREQS" => 5);
98 # Gathering and indexing split to areas
99 UnSet("CONFIG_AREAS");
102 UnSet("CONFIG_SITES");
104 # Gathering of images
105 UnSet("CONFIG_IMAGES");
107 # Gathering of MS Word files
108 UnSet("CONFIG_MSWORD");
110 # Gathering of MS Excel files
111 UnSet("CONFIG_EXCEL");
113 # Extra stuff for Centrum
114 UnSet("CONFIG_CENTRUM");
116 # Configuration for Morfeo
117 UnSet("CONFIG_MORFEO");
119 # Stemmer by UFAL and its dictionaries
120 UnSet("CONFIG_UFAL_STEMMER");
121 UnSet("CONFIG_UFAL_DICT_CS");
122 UnSet("CONFIG_UFAL_DICT_SK");
124 # Stemmer for Polish with its dictionary
125 UnSet("CONFIG_PL_STEMMER");
128 UnSet("CONFIG_WEIGHTS");
132 ### These are internal settings which shouldn't be modified by custom configs ###
134 # Default configuration file
135 Set("DEFAULT_CONFIG" => "cf/sherlock");
137 # Configuration of parts of the UCW library
138 Set("CONFIG_UCW_PERL" => 1);
139 Set("CONFIG_SHELL_UTILS" => 1);
141 # Configuration of the language library
142 Set("CONFIG_LANG_DETECT" => 1);
144 # We always want to be able to use large files
145 Set("CONFIG_LFS" => 1);
147 # Maximum word length common for Sherlock and liblang, see comments in sherlock/index.h
148 Set("MAX_WORD_LEN" => 64);