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 ###
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_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
40 UnSet("CONFIG_REAPD");
47 ### Optional Modules & Features ###
49 # Support databases >4GB
50 Set("CONFIG_LARGE_DB");
52 # Bucket position alignment shift (maximum bucket file size is 4G << this shift)
53 Set("CONFIG_BUCKET_SHIFT" => 7);
55 # Number of sections we remember limits for
56 Set("SHERLOCK_NUM_SECTIONS" => 4);
58 # Use shared libraries
59 UnSet("CONFIG_SHARED");
61 # Gathering of PDF files
64 # Searching by last modification time
65 Set("CONFIG_LASTMOD");
67 # Searching by file type
68 Set("CONFIG_FILETYPE");
70 # Language processing (requires CONFIG_FILETYPE)
76 # Enable indexing of context-dependent words and set maximum number
77 # of context slots (undef, 256 or 32768). Lower values save memory, higher time.
78 Set("CONFIG_MAX_CONTEXTS" => 256);
80 # Enable more than 16-bit word positions in references
81 Set("CONFIG_32BIT_REFERENCES");
84 UnSet("CONFIG_DEBUG_TOOLS");
86 # Search server: Enable EXPLAIN command (slows searching down slightly)
87 Set("CONFIG_EXPLAIN");
89 # Search server: Full support for all types of boolean expressions and ANY.
90 # (Otherwise, only expressions with at least one term matched positively are allowed.)
91 # It may be slow in some cases - for example `ANY' or `NOT "A"' queries.
92 UnSet("CONFIG_ALLOW_ANY");
94 # If your system doesn't contain GNU libc 2.3 or newer, it's recommended to let Sherlock
95 # use its own regex library (a copy of the glibc one), because the default regex library
96 # is likely to be crappy.
97 Set("CONFIG_OWN_REGEX");
99 # Include custom builtin functions in filtering language
100 UnSet("CONFIG_CUSTOM_FILTER");
102 # Install libraries and their API includes
103 UnSet("CONFIG_INSTALL_API");
106 # Number of refresh frequencies we allow
107 Set("SHERLOCK_NUM_FREQS" => 5);
109 # Gathering and indexing split to areas
110 UnSet("CONFIG_AREAS");
113 UnSet("CONFIG_SITES");
115 # Gathering of images
116 UnSet("CONFIG_IMAGES");
118 # Gathering of MS Word files
119 UnSet("CONFIG_MSWORD");
121 # Gathering of MS Excel files
122 UnSet("CONFIG_EXCEL");
124 # Extra stuff for Centrum
125 UnSet("CONFIG_CENTRUM");
126 UnSet("CONFIG_CENTRUM_CZ");
127 UnSet("CONFIG_CENTRUM_SK");
128 UnSet("CONFIG_CENTRUM_PL");
129 UnSet("CONFIG_CENTRUM_CATALOG");
131 # Debugging version of Centrum installation (lower limits, trivial filters etc.)
132 UnSet("CONFIG_CENTRUM_DEBUG");
134 # Stemmer by UFAL and its dictionaries
135 UnSet("CONFIG_UFAL_STEMMER");
136 UnSet("CONFIG_UFAL_DICT_CS");
137 UnSet("CONFIG_UFAL_DICT_SK");
139 # Stemmer for Polish with its dictionary
140 UnSet("CONFIG_PL_STEMMER");
143 UnSet("CONFIG_WEIGHTS");
145 # Include the incremental indexer
146 UnSet("CONFIG_INCREMENTAL");
148 # Maximum number of search servers, databases per search server and route keys
149 Set("MUX_MAX_SERVERS" => 128);
150 Set("MUX_MAX_DATABASES_PER_SERVER" => 4);
151 Set("MUX_MAX_ROUTE_KEYS" => 4);
155 ### Features which are decided on automatically (see sherlock/autoconf.cfg), but can be overriden ###
157 # The gatherer library
158 UnSet("CONFIG_GATHERER");
160 # The analyser library
161 UnSet("CONFIG_ANALYSER");
163 ### These are internal settings which shouldn't be modified by custom configs ###
165 # Default configuration file
166 Set("DEFAULT_CONFIG" => "cf/sherlock");
168 # Configuration of parts of the UCW library
169 Set("CONFIG_UCW_PERL" => 1);
170 Set("CONFIG_UCW_SHELL_UTILS" => 1);
172 # Configuration of the language library
173 Set("CONFIG_LANG_DETECT" => 1);
175 # We always want to be able to use large files
176 Set("CONFIG_LFS" => 1);
178 # Maximum word length common for Sherlock and liblang, see comments in sherlock/index.h
179 Set("MAX_WORD_LEN" => 64);