1
0
mirror of https://github.com/postgres/postgres.git synced 2025-07-07 00:36:50 +03:00

Teach parse_coerce about non-cachable functions (actually,

make it call eval_const_expressions() so that it doesn't have to know).
This commit is contained in:
Tom Lane
1999-10-02 23:29:19 +00:00
parent 6eb8d255d2
commit b40ac42efb
2 changed files with 18 additions and 33 deletions

View File

@ -7,13 +7,14 @@
*
*
* IDENTIFICATION
* $Header: /cvsroot/pgsql/src/backend/parser/parse_coerce.c,v 2.23 1999/08/24 00:09:56 tgl Exp $
* $Header: /cvsroot/pgsql/src/backend/parser/parse_coerce.c,v 2.24 1999/10/02 23:29:12 tgl Exp $
*
*-------------------------------------------------------------------------
*/
#include "postgres.h"
#include "optimizer/clauses.h"
#include "parser/parse_coerce.h"
#include "parser/parse_expr.h"
#include "parser/parse_func.h"
@ -49,7 +50,8 @@ coerce_type(ParseState *pstate, Node *node, Oid inputTypeId, Oid targetTypeId,
}
else if (inputTypeId == UNKNOWNOID && IsA(node, Const))
{
/* Input is a string constant with previously undetermined type.
/*
* Input is a string constant with previously undetermined type.
* Apply the target type's typinput function to it to produce
* a constant of the target type.
*
@ -58,6 +60,11 @@ coerce_type(ParseState *pstate, Node *node, Oid inputTypeId, Oid targetTypeId,
* necessarily behave the same as a type conversion function.
* For example, int4's typinput function will reject "1.2",
* whereas float-to-int type conversion will round to integer.
*
* XXX if the typinput function is not cachable, we really ought
* to postpone evaluation of the function call until runtime.
* But there is no way to represent a typinput function call as
* an expression tree, because C-string values are not Datums.
*/
Const *con = (Const *) node;
Const *newcon = makeNode(Const);
@ -97,38 +104,16 @@ coerce_type(ParseState *pstate, Node *node, Oid inputTypeId, Oid targetTypeId,
/*
* If the input is a constant, apply the type conversion function
* now instead of delaying to runtime. (This could someday be
* done in a downstream constant-expression-simplifier, but we
* can save cycles in the rewriter if we do it here.)
* now instead of delaying to runtime. (We could, of course,
* just leave this to be done during planning/optimization;
* but it's a very frequent special case, and we save cycles
* in the rewriter if we fold the expression now.)
*
* XXX there are cases where we probably shouldn't do this,
* such as coercing text 'now' to datetime? Need a way to
* know whether type conversion function is cacheable...
* Note that no folding will occur if the conversion function is
* not marked 'iscachable'.
*/
if (IsA(node, Const) && ! ((Const *) node)->constisnull)
{
Const *con = (Const *) node;
Oid convertFuncid;
Datum val;
Assert(IsA(result, Expr) &&
((Expr *) result)->opType == FUNC_EXPR);
/* Convert the given constant */
convertFuncid = ((Func *) (((Expr *) result)->oper))->funcid;
val = (Datum) fmgr(convertFuncid, con->constvalue);
/* now make a new const node */
con = makeNode(Const);
con->consttype = targetTypeId;
con->constlen = typeLen(targetType);
con->constbyval = typeByVal(targetType);
con->constvalue = val;
con->constisnull = false;
con->constisset = false;
result = (Node *) con;
}
if (IsA(node, Const))
result = eval_const_expressions(result);
}
return result;