version 1, including all changes.
.
Rev |
Author |
# |
Line |
1 |
perry |
1 |
HASH |
|
|
2 |
!!!HASH |
|
|
3 |
NAME |
|
|
4 |
SYNOPSIS |
|
|
5 |
DESCRIPTION |
|
|
6 |
ERRORS |
|
|
7 |
SEE ALSO |
|
|
8 |
BUGS |
|
|
9 |
---- |
|
|
10 |
!!NAME |
|
|
11 |
|
|
|
12 |
|
|
|
13 |
hash - hash database access method |
|
|
14 |
!!SYNOPSIS |
|
|
15 |
|
|
|
16 |
|
|
|
17 |
__#include |
|
|
18 |
__ |
|
|
19 |
!!DESCRIPTION |
|
|
20 |
|
|
|
21 |
|
|
|
22 |
The routine ''dbopen'' is the library interface to |
|
|
23 |
database files. One of the supported file formats is hash |
|
|
24 |
files. The general description of the database access |
|
|
25 |
methods is in dbopen(3), this manual page describes |
|
|
26 |
only the hash specific information. |
|
|
27 |
|
|
|
28 |
|
|
|
29 |
The hash data structure is an extensible, dynamic hashing |
|
|
30 |
scheme. |
|
|
31 |
|
|
|
32 |
|
|
|
33 |
The access method specific data structure provided to |
|
|
34 |
''dbopen'' is defined in the |
|
|
35 |
'' |
|
|
36 |
|
|
|
37 |
|
|
|
38 |
typedef struct { |
|
|
39 |
|
|
|
40 |
|
|
|
41 |
u_int bsize; |
|
|
42 |
u_int ffactor; |
|
|
43 |
u_int nelem; |
|
|
44 |
u_int cachesize; |
|
|
45 |
u_int32_t (*hash)(const void *, size_t); |
|
|
46 |
int lorder; |
|
|
47 |
|
|
|
48 |
|
|
|
49 |
} HASHINFO; |
|
|
50 |
|
|
|
51 |
|
|
|
52 |
The elements of this structure are as follows: |
|
|
53 |
|
|
|
54 |
|
|
|
55 |
bsize |
|
|
56 |
|
|
|
57 |
|
|
|
58 |
''Bsize'' defines the hash table bucket size, and is, by |
|
|
59 |
default, 256 bytes. It may be preferable to increase the |
|
|
60 |
page size for disk-resident tables and tables with large |
|
|
61 |
data items. |
|
|
62 |
|
|
|
63 |
|
|
|
64 |
ffactor |
|
|
65 |
|
|
|
66 |
|
|
|
67 |
''Ffactor'' indicates a desired density within the hash |
|
|
68 |
table. It is an approximation of the number of keys allowed |
|
|
69 |
to accumulate in any one bucket, determining when the hash |
|
|
70 |
table grows or shrinks. The default value is 8. |
|
|
71 |
|
|
|
72 |
|
|
|
73 |
nelem |
|
|
74 |
|
|
|
75 |
|
|
|
76 |
''Nelem'' is an estimate of the final size of the hash |
|
|
77 |
table. If not set or set too low, hash tables will expand |
|
|
78 |
gracefully as keys are entered, although a slight |
|
|
79 |
performance degradation may be noticed. The default value is |
|
|
80 |
1. |
|
|
81 |
|
|
|
82 |
|
|
|
83 |
cachesize |
|
|
84 |
|
|
|
85 |
|
|
|
86 |
A suggested maximum size, in bytes, of the memory cache. |
|
|
87 |
This value is __only__ advisory, and the access method |
|
|
88 |
will allocate more memory rather than fail. |
|
|
89 |
|
|
|
90 |
|
|
|
91 |
hash |
|
|
92 |
|
|
|
93 |
|
|
|
94 |
''Hash'' is a user defined hash function. Since no hash |
|
|
95 |
function performs equally well on all possible data, the |
|
|
96 |
user may find that the built-in hash function does poorly on |
|
|
97 |
a particular data set. User specified hash functions must |
|
|
98 |
take two arguments (a pointer to a byte string and a length) |
|
|
99 |
and return a 32-bit quantity to be used as the hash |
|
|
100 |
value. |
|
|
101 |
|
|
|
102 |
|
|
|
103 |
lorder |
|
|
104 |
|
|
|
105 |
|
|
|
106 |
The byte order for integers in the stored database metadata. |
|
|
107 |
The number should represent the order as an integer; for |
|
|
108 |
example, big endian order would be the number 4,321. If |
|
|
109 |
''lorder'' is 0 (no order is specified) the current host |
|
|
110 |
order is used. If the file already exists, the specified |
|
|
111 |
value is ignored and the value specified when the tree was |
|
|
112 |
created is used. |
|
|
113 |
|
|
|
114 |
|
|
|
115 |
If the file already exists (and the O_TRUNC flag is not |
|
|
116 |
specified), the values specified for the parameters bsize, |
|
|
117 |
ffactor, lorder and nelem are ignored and the values |
|
|
118 |
specified when the tree was created are used. |
|
|
119 |
|
|
|
120 |
|
|
|
121 |
If a hash function is specified, ''hash_open'' will |
|
|
122 |
attempt to determine if the hash function specified is the |
|
|
123 |
same as the one with which the database was created, and |
|
|
124 |
will fail if it is not. |
|
|
125 |
|
|
|
126 |
|
|
|
127 |
Backward compatible interfaces to the routines described in |
|
|
128 |
dbm(3), and ndbm(3) are provided, however |
|
|
129 |
these interfaces are not compatible with previous file |
|
|
130 |
formats. |
|
|
131 |
!!ERRORS |
|
|
132 |
|
|
|
133 |
|
|
|
134 |
The ''hash'' access method routines may fail and set |
|
|
135 |
''errno'' for any of the errors specified for the library |
|
|
136 |
routine dbopen(3). |
|
|
137 |
!!SEE ALSO |
|
|
138 |
|
|
|
139 |
|
|
|
140 |
btree(3), dbopen(3), mpool(3), |
|
|
141 |
recno(3) |
|
|
142 |
|
|
|
143 |
|
|
|
144 |
''Dynamic Hash Tables'', Per-Ake Larson, Communications |
|
|
145 |
of the ACM, April 1988. |
|
|
146 |
|
|
|
147 |
|
|
|
148 |
''A New Hash Package for UNIX'', Margo Seltzer, USENIX |
|
|
149 |
Proceedings, Winter 1991. |
|
|
150 |
!!BUGS |
|
|
151 |
|
|
|
152 |
|
|
|
153 |
Only big and little endian byte order is |
|
|
154 |
supported. |
|
|
155 |
---- |