<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /><title>64.3. Extensibility</title><link rel="stylesheet" type="text/css" href="stylesheet.css" /><link rev="made" href="pgsql-docs@lists.postgresql.org" /><meta name="generator" content="DocBook XSL Stylesheets Vsnapshot" /><link rel="prev" href="gist-builtin-opclasses.html" title="64.2. Built-in Operator Classes" /><link rel="next" href="gist-implementation.html" title="64.4. Implementation" /></head><body><div xmlns="http://www.w3.org/TR/xhtml1/transitional" class="navheader"><table width="100%" summary="Navigation header"><tr><th colspan="5" align="center">64.3. Extensibility</th></tr><tr><td width="10%" align="left"><a accesskey="p" href="gist-builtin-opclasses.html" title="64.2. Built-in Operator Classes">Prev</a> </td><td width="10%" align="left"><a accesskey="u" href="gist.html" title="Chapter 64. GiST Indexes">Up</a></td><th width="60%" align="center">Chapter 64. GiST Indexes</th><td width="10%" align="right"><a accesskey="h" href="index.html" title="PostgreSQL 11.5 Documentation">Home</a></td><td width="10%" align="right"> <a accesskey="n" href="gist-implementation.html" title="64.4. Implementation">Next</a></td></tr></table><hr></hr></div><div class="sect1" id="GIST-EXTENSIBILITY"><div class="titlepage"><div><div><h2 class="title" style="clear: both">64.3. Extensibility</h2></div></div></div><p>
Traditionally, implementing a new index access method meant a lot of
difficult work. It was necessary to understand the inner workings of the
database, such as the lock manager and Write-Ahead Log. The
<acronym class="acronym">GiST</acronym> interface has a high level of abstraction,
requiring the access method implementer only to implement the semantics of
the data type being accessed. The <acronym class="acronym">GiST</acronym> layer itself
takes care of concurrency, logging and searching the tree structure.
</p><p>
This extensibility should not be confused with the extensibility of the
other standard search trees in terms of the data they can handle. For
example, <span class="productname">PostgreSQL</span> supports extensible B-trees
and hash indexes. That means that you can use
<span class="productname">PostgreSQL</span> to build a B-tree or hash over any
data type you want. But B-trees only support range predicates
(<code class="literal"><</code>, <code class="literal">=</code>, <code class="literal">></code>),
and hash indexes only support equality queries.
</p><p>
So if you index, say, an image collection with a
<span class="productname">PostgreSQL</span> B-tree, you can only issue queries
such as <span class="quote">“<span class="quote">is imagex equal to imagey</span>”</span>, <span class="quote">“<span class="quote">is imagex less
than imagey</span>”</span> and <span class="quote">“<span class="quote">is imagex greater than imagey</span>”</span>.
Depending on how you define <span class="quote">“<span class="quote">equals</span>”</span>, <span class="quote">“<span class="quote">less than</span>”</span>
and <span class="quote">“<span class="quote">greater than</span>”</span> in this context, this could be useful.
However, by using a <acronym class="acronym">GiST</acronym> based index, you could create
ways to ask domain-specific questions, perhaps <span class="quote">“<span class="quote">find all images of
horses</span>”</span> or <span class="quote">“<span class="quote">find all over-exposed images</span>”</span>.
</p><p>
All it takes to get a <acronym class="acronym">GiST</acronym> access method up and running
is to implement several user-defined methods, which define the behavior of
keys in the tree. Of course these methods have to be pretty fancy to
support fancy queries, but for all the standard queries (B-trees,
R-trees, etc.) they're relatively straightforward. In short,
<acronym class="acronym">GiST</acronym> combines extensibility along with generality, code
reuse, and a clean interface.
</p><p>
There are five methods that an index operator class for
<acronym class="acronym">GiST</acronym> must provide, and four that are optional.
Correctness of the index is ensured
by proper implementation of the <code class="function">same</code>, <code class="function">consistent</code>
and <code class="function">union</code> methods, while efficiency (size and speed) of the
index will depend on the <code class="function">penalty</code> and <code class="function">picksplit</code>
methods.
Two optional methods are <code class="function">compress</code> and
<code class="function">decompress</code>, which allow an index to have internal tree data of
a different type than the data it indexes. The leaves are to be of the
indexed data type, while the other tree nodes can be of any C struct (but
you still have to follow <span class="productname">PostgreSQL</span> data type rules here,
see about <code class="literal">varlena</code> for variable sized data). If the tree's
internal data type exists at the SQL level, the <code class="literal">STORAGE</code> option
of the <code class="command">CREATE OPERATOR CLASS</code> command can be used.
The optional eighth method is <code class="function">distance</code>, which is needed
if the operator class wishes to support ordered scans (nearest-neighbor
searches). The optional ninth method <code class="function">fetch</code> is needed if the
operator class wishes to support index-only scans, except when the
<code class="function">compress</code> method is omitted.
</p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><code class="function">consistent</code></span></dt><dd><p>
Given an index entry <code class="literal">p</code> and a query value <code class="literal">q</code>,
this function determines whether the index entry is
<span class="quote">“<span class="quote">consistent</span>”</span> with the query; that is, could the predicate
<span class="quote">“<span class="quote"><em class="replaceable"><code>indexed_column</code></em>
<em class="replaceable"><code>indexable_operator</code></em> <code class="literal">q</code></span>”</span> be true for
any row represented by the index entry? For a leaf index entry this is
equivalent to testing the indexable condition, while for an internal
tree node this determines whether it is necessary to scan the subtree
of the index represented by the tree node. When the result is
<code class="literal">true</code>, a <code class="literal">recheck</code> flag must also be returned.
This indicates whether the predicate is certainly true or only possibly
true. If <code class="literal">recheck</code> = <code class="literal">false</code> then the index has
tested the predicate condition exactly, whereas if <code class="literal">recheck</code>
= <code class="literal">true</code> the row is only a candidate match. In that case the
system will automatically evaluate the
<em class="replaceable"><code>indexable_operator</code></em> against the actual row value to see
if it is really a match. This convention allows
<acronym class="acronym">GiST</acronym> to support both lossless and lossy index
structures.
</p><p>
The <acronym class="acronym">SQL</acronym> declaration of the function must look like this:
</p><pre class="programlisting">
CREATE OR REPLACE FUNCTION my_consistent(internal, data_type, smallint, oid, internal)
RETURNS bool
AS 'MODULE_PATHNAME'
LANGUAGE C STRICT;
</pre><p>
And the matching code in the C module could then follow this skeleton:
</p><pre class="programlisting">
PG_FUNCTION_INFO_V1(my_consistent);
Datum
my_consistent(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
data_type *query = PG_GETARG_DATA_TYPE_P(1);
StrategyNumber strategy = (StrategyNumber) PG_GETARG_UINT16(2);
/* Oid subtype = PG_GETARG_OID(3); */
bool *recheck = (bool *) PG_GETARG_POINTER(4);
data_type *key = DatumGetDataType(entry->key);
bool retval;
/*
* determine return value as a function of strategy, key and query.
*
* Use GIST_LEAF(entry) to know where you're called in the index tree,
* which comes handy when supporting the = operator for example (you could
* check for non empty union() in non-leaf nodes and equality in leaf
* nodes).
*/
*recheck = true; /* or false if check is exact */
PG_RETURN_BOOL(retval);
}
</pre><p>
Here, <code class="varname">key</code> is an element in the index and <code class="varname">query</code>
the value being looked up in the index. The <code class="literal">StrategyNumber</code>
parameter indicates which operator of your operator class is being
applied — it matches one of the operator numbers in the
<code class="command">CREATE OPERATOR CLASS</code> command.
</p><p>
Depending on which operators you have included in the class, the data
type of <code class="varname">query</code> could vary with the operator, since it will
be whatever type is on the righthand side of the operator, which might
be different from the indexed data type appearing on the lefthand side.
(The above code skeleton assumes that only one type is possible; if
not, fetching the <code class="varname">query</code> argument value would have to depend
on the operator.) It is recommended that the SQL declaration of
the <code class="function">consistent</code> function use the opclass's indexed data
type for the <code class="varname">query</code> argument, even though the actual type
might be something else depending on the operator.
</p></dd><dt><span class="term"><code class="function">union</code></span></dt><dd><p>
This method consolidates information in the tree. Given a set of
entries, this function generates a new index entry that represents
all the given entries.
</p><p>
The <acronym class="acronym">SQL</acronym> declaration of the function must look like this:
</p><pre class="programlisting">
CREATE OR REPLACE FUNCTION my_union(internal, internal)
RETURNS storage_type
AS 'MODULE_PATHNAME'
LANGUAGE C STRICT;
</pre><p>
And the matching code in the C module could then follow this skeleton:
</p><pre class="programlisting">
PG_FUNCTION_INFO_V1(my_union);
Datum
my_union(PG_FUNCTION_ARGS)
{
GistEntryVector *entryvec = (GistEntryVector *) PG_GETARG_POINTER(0);
GISTENTRY *ent = entryvec->vector;
data_type *out,
*tmp,
*old;
int numranges,
i = 0;
numranges = entryvec->n;
tmp = DatumGetDataType(ent[0].key);
out = tmp;
if (numranges == 1)
{
out = data_type_deep_copy(tmp);
PG_RETURN_DATA_TYPE_P(out);
}
for (i = 1; i < numranges; i++)
{
old = out;
tmp = DatumGetDataType(ent[i].key);
out = my_union_implementation(out, tmp);
}
PG_RETURN_DATA_TYPE_P(out);
}
</pre><p>
</p><p>
As you can see, in this skeleton we're dealing with a data type
where <code class="literal">union(X, Y, Z) = union(union(X, Y), Z)</code>. It's easy
enough to support data types where this is not the case, by
implementing the proper union algorithm in this
<acronym class="acronym">GiST</acronym> support method.
</p><p>
The result of the <code class="function">union</code> function must be a value of the
index's storage type, whatever that is (it might or might not be
different from the indexed column's type). The <code class="function">union</code>
function should return a pointer to newly <code class="function">palloc()</code>ed
memory. You can't just return the input value as-is, even if there is
no type change.
</p><p>
As shown above, the <code class="function">union</code> function's
first <code class="type">internal</code> argument is actually
a <code class="structname">GistEntryVector</code> pointer. The second argument is a
pointer to an integer variable, which can be ignored. (It used to be
required that the <code class="function">union</code> function store the size of its
result value into that variable, but this is no longer necessary.)
</p></dd><dt><span class="term"><code class="function">compress</code></span></dt><dd><p>
Converts a data item into a format suitable for physical storage in
an index page.
If the <code class="function">compress</code> method is omitted, data items are stored
in the index without modification.
</p><p>
The <acronym class="acronym">SQL</acronym> declaration of the function must look like this:
</p><pre class="programlisting">
CREATE OR REPLACE FUNCTION my_compress(internal)
RETURNS internal
AS 'MODULE_PATHNAME'
LANGUAGE C STRICT;
</pre><p>
And the matching code in the C module could then follow this skeleton:
</p><pre class="programlisting">
PG_FUNCTION_INFO_V1(my_compress);
Datum
my_compress(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
GISTENTRY *retval;
if (entry->leafkey)
{
/* replace entry->key with a compressed version */
compressed_data_type *compressed_data = palloc(sizeof(compressed_data_type));
/* fill *compressed_data from entry->key ... */
retval = palloc(sizeof(GISTENTRY));
gistentryinit(*retval, PointerGetDatum(compressed_data),
entry->rel, entry->page, entry->offset, FALSE);
}
else
{
/* typically we needn't do anything with non-leaf entries */
retval = entry;
}
PG_RETURN_POINTER(retval);
}
</pre><p>
</p><p>
You have to adapt <em class="replaceable"><code>compressed_data_type</code></em> to the specific
type you're converting to in order to compress your leaf nodes, of
course.
</p></dd><dt><span class="term"><code class="function">decompress</code></span></dt><dd><p>
Converts the stored representation of a data item into a format that
can be manipulated by the other GiST methods in the operator class.
If the <code class="function">decompress</code> method is omitted, it is assumed that
the other GiST methods can work directly on the stored data format.
(<code class="function">decompress</code> is not necessarily the reverse of
the <code class="function">compress</code> method; in particular,
if <code class="function">compress</code> is lossy then it's impossible
for <code class="function">decompress</code> to exactly reconstruct the original
data. <code class="function">decompress</code> is not necessarily equivalent
to <code class="function">fetch</code>, either, since the other GiST methods might not
require full reconstruction of the data.)
</p><p>
The <acronym class="acronym">SQL</acronym> declaration of the function must look like this:
</p><pre class="programlisting">
CREATE OR REPLACE FUNCTION my_decompress(internal)
RETURNS internal
AS 'MODULE_PATHNAME'
LANGUAGE C STRICT;
</pre><p>
And the matching code in the C module could then follow this skeleton:
</p><pre class="programlisting">
PG_FUNCTION_INFO_V1(my_decompress);
Datum
my_decompress(PG_FUNCTION_ARGS)
{
PG_RETURN_POINTER(PG_GETARG_POINTER(0));
}
</pre><p>
The above skeleton is suitable for the case where no decompression
is needed. (But, of course, omitting the method altogether is even
easier, and is recommended in such cases.)
</p></dd><dt><span class="term"><code class="function">penalty</code></span></dt><dd><p>
Returns a value indicating the <span class="quote">“<span class="quote">cost</span>”</span> of inserting the new
entry into a particular branch of the tree. Items will be inserted
down the path of least <code class="function">penalty</code> in the tree.
Values returned by <code class="function">penalty</code> should be non-negative.
If a negative value is returned, it will be treated as zero.
</p><p>
The <acronym class="acronym">SQL</acronym> declaration of the function must look like this:
</p><pre class="programlisting">
CREATE OR REPLACE FUNCTION my_penalty(internal, internal, internal)
RETURNS internal
AS 'MODULE_PATHNAME'
LANGUAGE C STRICT; -- in some cases penalty functions need not be strict
</pre><p>
And the matching code in the C module could then follow this skeleton:
</p><pre class="programlisting">
PG_FUNCTION_INFO_V1(my_penalty);
Datum
my_penalty(PG_FUNCTION_ARGS)
{
GISTENTRY *origentry = (GISTENTRY *) PG_GETARG_POINTER(0);
GISTENTRY *newentry = (GISTENTRY *) PG_GETARG_POINTER(1);
float *penalty = (float *) PG_GETARG_POINTER(2);
data_type *orig = DatumGetDataType(origentry->key);
data_type *new = DatumGetDataType(newentry->key);
*penalty = my_penalty_implementation(orig, new);
PG_RETURN_POINTER(penalty);
}
</pre><p>
For historical reasons, the <code class="function">penalty</code> function doesn't
just return a <code class="type">float</code> result; instead it has to store the value
at the location indicated by the third argument. The return
value per se is ignored, though it's conventional to pass back the
address of that argument.
</p><p>
The <code class="function">penalty</code> function is crucial to good performance of
the index. It'll get used at insertion time to determine which branch
to follow when choosing where to add the new entry in the tree. At
query time, the more balanced the index, the quicker the lookup.
</p></dd><dt><span class="term"><code class="function">picksplit</code></span></dt><dd><p>
When an index page split is necessary, this function decides which
entries on the page are to stay on the old page, and which are to move
to the new page.
</p><p>
The <acronym class="acronym">SQL</acronym> declaration of the function must look like this:
</p><pre class="programlisting">
CREATE OR REPLACE FUNCTION my_picksplit(internal, internal)
RETURNS internal
AS 'MODULE_PATHNAME'
LANGUAGE C STRICT;
</pre><p>
And the matching code in the C module could then follow this skeleton:
</p><pre class="programlisting">
PG_FUNCTION_INFO_V1(my_picksplit);
Datum
my_picksplit(PG_FUNCTION_ARGS)
{
GistEntryVector *entryvec = (GistEntryVector *) PG_GETARG_POINTER(0);
GIST_SPLITVEC *v = (GIST_SPLITVEC *) PG_GETARG_POINTER(1);
OffsetNumber maxoff = entryvec->n - 1;
GISTENTRY *ent = entryvec->vector;
int i,
nbytes;
OffsetNumber *left,
*right;
data_type *tmp_union;
data_type *unionL;
data_type *unionR;
GISTENTRY **raw_entryvec;
maxoff = entryvec->n - 1;
nbytes = (maxoff + 1) * sizeof(OffsetNumber);
v->spl_left = (OffsetNumber *) palloc(nbytes);
left = v->spl_left;
v->spl_nleft = 0;
v->spl_right = (OffsetNumber *) palloc(nbytes);
right = v->spl_right;
v->spl_nright = 0;
unionL = NULL;
unionR = NULL;
/* Initialize the raw entry vector. */
raw_entryvec = (GISTENTRY **) malloc(entryvec->n * sizeof(void *));
for (i = FirstOffsetNumber; i <= maxoff; i = OffsetNumberNext(i))
raw_entryvec[i] = &(entryvec->vector[i]);
for (i = FirstOffsetNumber; i <= maxoff; i = OffsetNumberNext(i))
{
int real_index = raw_entryvec[i] - entryvec->vector;
tmp_union = DatumGetDataType(entryvec->vector[real_index].key);
Assert(tmp_union != NULL);
/*
* Choose where to put the index entries and update unionL and unionR
* accordingly. Append the entries to either v_spl_left or
* v_spl_right, and care about the counters.
*/
if (my_choice_is_left(unionL, curl, unionR, curr))
{
if (unionL == NULL)
unionL = tmp_union;
else
unionL = my_union_implementation(unionL, tmp_union);
*left = real_index;
++left;
++(v->spl_nleft);
}
else
{
/*
* Same on the right
*/
}
}
v->spl_ldatum = DataTypeGetDatum(unionL);
v->spl_rdatum = DataTypeGetDatum(unionR);
PG_RETURN_POINTER(v);
}
</pre><p>
Notice that the <code class="function">picksplit</code> function's result is delivered
by modifying the passed-in <code class="structname">v</code> structure. The return
value per se is ignored, though it's conventional to pass back the
address of <code class="structname">v</code>.
</p><p>
Like <code class="function">penalty</code>, the <code class="function">picksplit</code> function
is crucial to good performance of the index. Designing suitable
<code class="function">penalty</code> and <code class="function">picksplit</code> implementations
is where the challenge of implementing well-performing
<acronym class="acronym">GiST</acronym> indexes lies.
</p></dd><dt><span class="term"><code class="function">same</code></span></dt><dd><p>
Returns true if two index entries are identical, false otherwise.
(An <span class="quote">“<span class="quote">index entry</span>”</span> is a value of the index's storage type,
not necessarily the original indexed column's type.)
</p><p>
The <acronym class="acronym">SQL</acronym> declaration of the function must look like this:
</p><pre class="programlisting">
CREATE OR REPLACE FUNCTION my_same(storage_type, storage_type, internal)
RETURNS internal
AS 'MODULE_PATHNAME'
LANGUAGE C STRICT;
</pre><p>
And the matching code in the C module could then follow this skeleton:
</p><pre class="programlisting">
PG_FUNCTION_INFO_V1(my_same);
Datum
my_same(PG_FUNCTION_ARGS)
{
prefix_range *v1 = PG_GETARG_PREFIX_RANGE_P(0);
prefix_range *v2 = PG_GETARG_PREFIX_RANGE_P(1);
bool *result = (bool *) PG_GETARG_POINTER(2);
*result = my_eq(v1, v2);
PG_RETURN_POINTER(result);
}
</pre><p>
For historical reasons, the <code class="function">same</code> function doesn't
just return a Boolean result; instead it has to store the flag
at the location indicated by the third argument. The return
value per se is ignored, though it's conventional to pass back the
address of that argument.
</p></dd><dt><span class="term"><code class="function">distance</code></span></dt><dd><p>
Given an index entry <code class="literal">p</code> and a query value <code class="literal">q</code>,
this function determines the index entry's
<span class="quote">“<span class="quote">distance</span>”</span> from the query value. This function must be
supplied if the operator class contains any ordering operators.
A query using the ordering operator will be implemented by returning
index entries with the smallest <span class="quote">“<span class="quote">distance</span>”</span> values first,
so the results must be consistent with the operator's semantics.
For a leaf index entry the result just represents the distance to
the index entry; for an internal tree node, the result must be the
smallest distance that any child entry could have.
</p><p>
The <acronym class="acronym">SQL</acronym> declaration of the function must look like this:
</p><pre class="programlisting">
CREATE OR REPLACE FUNCTION my_distance(internal, data_type, smallint, oid, internal)
RETURNS float8
AS 'MODULE_PATHNAME'
LANGUAGE C STRICT;
</pre><p>
And the matching code in the C module could then follow this skeleton:
</p><pre class="programlisting">
PG_FUNCTION_INFO_V1(my_distance);
Datum
my_distance(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
data_type *query = PG_GETARG_DATA_TYPE_P(1);
StrategyNumber strategy = (StrategyNumber) PG_GETARG_UINT16(2);
/* Oid subtype = PG_GETARG_OID(3); */
/* bool *recheck = (bool *) PG_GETARG_POINTER(4); */
data_type *key = DatumGetDataType(entry->key);
double retval;
/*
* determine return value as a function of strategy, key and query.
*/
PG_RETURN_FLOAT8(retval);
}
</pre><p>
The arguments to the <code class="function">distance</code> function are identical to
the arguments of the <code class="function">consistent</code> function.
</p><p>
Some approximation is allowed when determining the distance, so long
as the result is never greater than the entry's actual distance. Thus,
for example, distance to a bounding box is usually sufficient in
geometric applications. For an internal tree node, the distance
returned must not be greater than the distance to any of the child
nodes. If the returned distance is not exact, the function must set
<code class="literal">*recheck</code> to true. (This is not necessary for internal tree
nodes; for them, the calculation is always assumed to be inexact.) In
this case the executor will calculate the accurate distance after
fetching the tuple from the heap, and reorder the tuples if necessary.
</p><p>
If the distance function returns <code class="literal">*recheck = true</code> for any
leaf node, the original ordering operator's return type must
be <code class="type">float8</code> or <code class="type">float4</code>, and the distance function's
result values must be comparable to those of the original ordering
operator, since the executor will sort using both distance function
results and recalculated ordering-operator results. Otherwise, the
distance function's result values can be any finite <code class="type">float8</code>
values, so long as the relative order of the result values matches the
order returned by the ordering operator. (Infinity and minus infinity
are used internally to handle cases such as nulls, so it is not
recommended that <code class="function">distance</code> functions return these values.)
</p></dd><dt><span class="term"><code class="function">fetch</code></span></dt><dd><p>
Converts the compressed index representation of a data item into the
original data type, for index-only scans. The returned data must be an
exact, non-lossy copy of the originally indexed value.
</p><p>
The <acronym class="acronym">SQL</acronym> declaration of the function must look like this:
</p><pre class="programlisting">
CREATE OR REPLACE FUNCTION my_fetch(internal)
RETURNS internal
AS 'MODULE_PATHNAME'
LANGUAGE C STRICT;
</pre><p>
The argument is a pointer to a <code class="structname">GISTENTRY</code> struct. On
entry, its <code class="structfield">key</code> field contains a non-NULL leaf datum in
compressed form. The return value is another <code class="structname">GISTENTRY</code>
struct, whose <code class="structfield">key</code> field contains the same datum in its
original, uncompressed form. If the opclass's compress function does
nothing for leaf entries, the <code class="function">fetch</code> method can return the
argument as-is. Or, if the opclass does not have a compress function,
the <code class="function">fetch</code> method can be omitted as well, since it would
necessarily be a no-op.
</p><p>
The matching code in the C module could then follow this skeleton:
</p><pre class="programlisting">
PG_FUNCTION_INFO_V1(my_fetch);
Datum
my_fetch(PG_FUNCTION_ARGS)
{
GISTENTRY *entry = (GISTENTRY *) PG_GETARG_POINTER(0);
input_data_type *in = DatumGetP(entry->key);
fetched_data_type *fetched_data;
GISTENTRY *retval;
retval = palloc(sizeof(GISTENTRY));
fetched_data = palloc(sizeof(fetched_data_type));
/*
* Convert 'fetched_data' into the a Datum of the original datatype.
*/
/* fill *retval from fetch_data. */
gistentryinit(*retval, PointerGetDatum(converted_datum),
entry->rel, entry->page, entry->offset, FALSE);
PG_RETURN_POINTER(retval);
}
</pre><p>
</p><p>
If the compress method is lossy for leaf entries, the operator class
cannot support index-only scans, and must not define
a <code class="function">fetch</code> function.
</p></dd></dl></div><p>
All the GiST support methods are normally called in short-lived memory
contexts; that is, <code class="varname">CurrentMemoryContext</code> will get reset after
each tuple is processed. It is therefore not very important to worry about
pfree'ing everything you palloc. However, in some cases it's useful for a
support method to cache data across repeated calls. To do that, allocate
the longer-lived data in <code class="literal">fcinfo->flinfo->fn_mcxt</code>, and
keep a pointer to it in <code class="literal">fcinfo->flinfo->fn_extra</code>. Such
data will survive for the life of the index operation (e.g., a single GiST
index scan, index build, or index tuple insertion). Be careful to pfree
the previous value when replacing a <code class="literal">fn_extra</code> value, or the leak
will accumulate for the duration of the operation.
</p></div><div class="navfooter"><hr /><table width="100%" summary="Navigation footer"><tr><td width="40%" align="left"><a accesskey="p" href="gist-builtin-opclasses.html">Prev</a> </td><td width="20%" align="center"><a accesskey="u" href="gist.html">Up</a></td><td width="40%" align="right"> <a accesskey="n" href="gist-implementation.html">Next</a></td></tr><tr><td width="40%" align="left" valign="top">64.2. Built-in Operator Classes </td><td width="20%" align="center"><a accesskey="h" href="index.html">Home</a></td><td width="40%" align="right" valign="top"> 64.4. Implementation</td></tr></table></div></body></html>
