Next: Arbitrary Octave types, Previous: Concept, Up: Top [Index]
For types specific to database implementations, see Postgresql data types.
The last column indicates whether the SQL type is automatically deduced by this packages functions from the representation of the data in Octave. If not, the correct SQL type must be specified by the user if the data is sent from Octave to the database.
The listed names of SQL data types are those which are used by the
functions of this package. They are sometimes different from the
original SQL typenames, e.g. int2, int4, and int8
instead of smallint, integer, and bigint,
respectively. If not obvious, the original SQL typename is given in
parantheses.
| SQL | Octave | auto |
|---|---|---|
| bool | logical scalar | yes |
| bytea (variable length binary string) | array of uint8, one-dimensional if converted from SQL data | yes |
| float8 | double scalar | yes |
| float4 | single scalar | yes |
| varchar | string | no |
| bpchar (char) | string | no |
| int2 | int16 scalar | yes |
| int4 | int32 scalar | yes |
| int8 | int64 scalar | yes |
| timestamp | 8-byte-time-value (see below), positive or negative difference to 2000-01-01 00:00 | no |
| timestamptz | as timestamp | no |
| time | 8-byte-time-value (see below) | no |
| timetz | 2-element cell array with 8-byte-time-value (see below, time of day) and int32 scalar (time zone in seconds, negative east of UTC) | no |
| date | int32 scalar, positive or negative difference to 2000-01-01 | no |
| interval | 3-element cell array with 8-byte-time-value (see below), int32 (days), and int32 (months) | no |
| bit | structure with fields bitlen (int32, number of valid
bits) and bits (uint8 array, 8 bits per entry, first entry
contains the leftmost bits, last entry may contain less than 8 bits) | no |
| varbit | as bit | yes |
| xml | string | no |
| any array | Structure with fields data (holding a cell-array with
entries of a type corresponding to the SQL element type), ndims
(holding the number of dimensions of the corresponding SQL array, since
this can not be deduced from the dimensions of the Octave cell-array in
all cases), and optionally (but always present in returned values)
lbounds (a row vector of enumeration bases for all dimensions,
default is ones (1, ndims); enumeration bases different from
1 are supported e.g. by Postgresql). Array elements may not
correspond to arrays in SQL (use additional dimensions for this), but
may correspond to composite types, which is allowed to lead to
arbitrarily deep nesting. | no |
| any composite type (row type) | One-dimensional cell-array with entries of types corresponding to the respective SQL types. Entries may also correspond to an array-type or composite type; this is allowed to lead to arbitrarily deep nesting. | no |
8-byte-time-value: int64 scalar, representing microseconds, if server is configured for integer date/time; double scalar, representing seconds, if server is configured for float date/time. If the type (integer or float) of an input Octave variable for an 8-byte-time-value does not match the server configuration, there is no automatic conversion but an error is thrown.
Octaves NA corresponds to an SQL NULL value (not NaN,
which is interpreted as a value of a float type!).
Next: Arbitrary Octave types, Previous: Concept, Up: Top [Index]