1 # Configuration variables of Sherlock and their default values
2 # (c) 2005--2006 Martin Mares <mj@ucw.cz>
4 # Version of the whole package
5 Set("SHERLOCK_VERSION" => "3.9");
7 # Installation directory for `make install'
8 Set("INSTALL_DIR" => "~/run-3.9");
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 # Support databases >4GB
53 Set("CONFIG_LARGE_DB");
55 # Bucket position alignment shift (maximum bucket file size is 4G << this shift)
56 Set("CONFIG_BUCKET_SHIFT" => 7);
58 # Number of sections we remember limits for
59 Set("SHERLOCK_NUM_SECTIONS" => 4);
61 # Use shared libraries
62 UnSet("CONFIG_SHARED");
64 # Gathering of PDF files
67 # Searching by last modification time
68 Set("CONFIG_LASTMOD");
70 # Searching by file type
71 Set("CONFIG_FILETYPE");
73 # Language processing (requires CONFIG_FILETYPE)
79 # Enable indexing of context-dependent words and set maximum number
80 # of context slots (undef, 256 or 32768). Lower values save memory, higher time.
81 Set("CONFIG_MAX_CONTEXTS" => 256);
83 # Enable more than 16-bit word positions in references
84 Set("CONFIG_32BIT_REFERENCES");
87 UnSet("CONFIG_DEBUG_TOOLS");
89 # Search server: Enable EXPLAIN command (slows searching down slightly)
90 Set("CONFIG_EXPLAIN");
92 # Search server: Full support for all types of boolean expressions and ANY.
93 # (Otherwise, only expressions with at least one term matched positively are allowed.)
94 # It may be slow in some cases - for example `ANY' or `NOT "A"' queries.
95 UnSet("CONFIG_ALLOW_ANY");
97 # If your system doesn't contain GNU libc 2.3 or newer, it's recommended to let Sherlock
98 # use its own regex library (a copy of the glibc one), because the default regex library
99 # is likely to be crappy.
100 Set("CONFIG_OWN_REGEX");
102 # Include custom builtin functions in filtering language
103 UnSet("CONFIG_CUSTOM_FILTER");
105 # Install libraries and their API includes
106 UnSet("CONFIG_INSTALL_API");
109 # Number of refresh frequencies we allow
110 Set("SHERLOCK_NUM_FREQS" => 5);
112 # Gathering and indexing split to areas
113 UnSet("CONFIG_AREAS");
116 UnSet("CONFIG_SITES");
118 # Gathering of images
119 UnSet("CONFIG_IMAGES");
121 # Gathering of MS Word files
122 UnSet("CONFIG_MSWORD");
124 # Gathering of MS Excel files
125 UnSet("CONFIG_EXCEL");
127 # Extra stuff for Centrum
128 UnSet("CONFIG_CENTRUM");
129 UnSet("CONFIG_CENTRUM_CZ");
130 UnSet("CONFIG_CENTRUM_SK");
131 UnSet("CONFIG_CENTRUM_PL");
132 UnSet("CONFIG_CENTRUM_CATALOG");
134 # Debugging version of Centrum installation (lower limits, trivial filters etc.)
135 UnSet("CONFIG_CENTRUM_DEBUG");
137 # Stemmer by UFAL and its dictionaries
138 UnSet("CONFIG_UFAL_STEMMER");
139 UnSet("CONFIG_UFAL_DICT_CS");
140 UnSet("CONFIG_UFAL_DICT_SK");
142 # Stemmer for Polish with its dictionary
143 UnSet("CONFIG_PL_STEMMER");
146 UnSet("CONFIG_WEIGHTS");
148 # Include the incremental indexer
149 UnSet("CONFIG_INCREMENTAL");
151 # Maximum number of search servers, databases per search server and route keys
152 Set("MUX_MAX_SERVERS" => 128);
153 Set("MUX_MAX_DATABASES_PER_SERVER" => 4);
154 Set("MUX_MAX_ROUTE_KEYS" => 4);
158 ### These are internal settings which shouldn't be modified by custom configs ###
160 # Default configuration file
161 Set("DEFAULT_CONFIG" => "cf/sherlock");
163 # Configuration of parts of the UCW library
164 Set("CONFIG_UCW_PERL" => 1);
165 Set("CONFIG_UCW_SHELL_UTILS" => 1);
167 # Configuration of the language library
168 Set("CONFIG_LANG_DETECT" => 1);
170 # We always want to be able to use large files
171 Set("CONFIG_LFS" => 1);
173 # Maximum word length common for Sherlock and liblang, see comments in sherlock/index.h
174 Set("MAX_WORD_LEN" => 64);