Re: [HACKERS] int2vector and btree indexes

2016-10-12 Thread Tom Lane
Amit Langote  writes:
> On 2016/10/11 21:40, Tom Lane wrote:
>> Hmm ... I kind of wonder why we have int2vectoreq or hashint2vector at
>> all, likewise the hash opclass based on them.

> Agreed.  So how about the attached patch to remove the said infrastructure?

Looks good, pushed.

regards, tom lane


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers


Re: [HACKERS] int2vector and btree indexes

2016-10-11 Thread Amit Langote
On 2016/10/11 21:40, Tom Lane wrote:
> Amit Langote  writes:
>> I was wrong that the index *never* gets used.  It does in fact get used if
>> the operator is an ordering search operator (<, <=, >, >=), in which case
>> the query would use an array_ops operator (which is a btree operator class
>> for type anyarray) and hence matches the index operator family.  I failed
>> to mention in my original message that int2vector_ops is a hash operator
>> class.  There is exactly one =(int2vector, int2vector) operator in the
>> system of which there is no btree equivalent.
> 
> Hmm ... I kind of wonder why we have int2vectoreq or hashint2vector at
> all, likewise the hash opclass based on them.  The code says that they
> are needed to support catcache index columns, but the only columns of
> this type are
> 
> regression=# select attrelid::regclass,attname from pg_attribute where 
> atttypid = 'int2vector'::regtype;
>   attrelid  |  attname  
> +---
>  pg_index   | indkey
>  pg_index   | indoption
>  pg_trigger | tgattr
> (3 rows)
> 
> and those don't have indexes at all, let alone catcaches based on them.
> So it looks to me like we could remove this infrastructure.  There is
> value in being able to hash int2vectors during queries, for sure, but
> we could let that be done by the anyarray hash opclass.

Agreed.  So how about the attached patch to remove the said infrastructure?

> Having said that, int2vector is not meant as a user-facing type and so
> I don't particularly care whether indexes built on it work conveniently.
> But it looks to me like we've got some unnecessary code here.

Ah, I did wonder whether int2vector has been deprecated as a user-facing
type.  Anyway after applying the patch, it seems that the original
complaint I raised is no longer an issue (or so I think) - operators
applied to int2vector are always resolved to those accepting anyarray and
matched with anyarray_ops of the correct index access method.

Thanks,
Amit
diff --git a/src/backend/access/hash/hashfunc.c b/src/backend/access/hash/hashfunc.c
index 614f4ff..12dce2e 100644
--- a/src/backend/access/hash/hashfunc.c
+++ b/src/backend/access/hash/hashfunc.c
@@ -131,14 +131,6 @@ hashoidvector(PG_FUNCTION_ARGS)
 }
 
 Datum
-hashint2vector(PG_FUNCTION_ARGS)
-{
-	int2vector *key = (int2vector *) PG_GETARG_POINTER(0);
-
-	return hash_any((unsigned char *) key->values, key->dim1 * sizeof(int16));
-}
-
-Datum
 hashname(PG_FUNCTION_ARGS)
 {
 	char	   *key = NameStr(*PG_GETARG_NAME(0));
diff --git a/src/backend/utils/adt/int.c b/src/backend/utils/adt/int.c
index 29d92a7..443ac5c 100644
--- a/src/backend/utils/adt/int.c
+++ b/src/backend/utils/adt/int.c
@@ -254,22 +254,6 @@ int2vectorsend(PG_FUNCTION_ARGS)
 	return array_send(fcinfo);
 }
 
-/*
- * We don't have a complete set of int2vector support routines,
- * but we need int2vectoreq for catcache indexing.
- */
-Datum
-int2vectoreq(PG_FUNCTION_ARGS)
-{
-	int2vector *a = (int2vector *) PG_GETARG_POINTER(0);
-	int2vector *b = (int2vector *) PG_GETARG_POINTER(1);
-
-	if (a->dim1 != b->dim1)
-		PG_RETURN_BOOL(false);
-	PG_RETURN_BOOL(memcmp(a->values, b->values, a->dim1 * sizeof(int16)) == 0);
-}
-
-
 /*
  *	 PUBLIC ROUTINES		 *
  */
diff --git a/src/backend/utils/cache/catcache.c b/src/backend/utils/cache/catcache.c
index db7099f..6016d19 100644
--- a/src/backend/utils/cache/catcache.c
+++ b/src/backend/utils/cache/catcache.c
@@ -126,11 +126,6 @@ GetCCHashEqFuncs(Oid keytype, PGFunction *hashfunc, RegProcedure *eqfunc)
 
 			*eqfunc = F_INT2EQ;
 			break;
-		case INT2VECTOROID:
-			*hashfunc = hashint2vector;
-
-			*eqfunc = F_INT2VECTOREQ;
-			break;
 		case INT4OID:
 			*hashfunc = hashint4;
 
diff --git a/src/include/access/hash.h b/src/include/access/hash.h
index 491d4c9..725e2f2 100644
--- a/src/include/access/hash.h
+++ b/src/include/access/hash.h
@@ -283,7 +283,6 @@ extern Datum hashenum(PG_FUNCTION_ARGS);
 extern Datum hashfloat4(PG_FUNCTION_ARGS);
 extern Datum hashfloat8(PG_FUNCTION_ARGS);
 extern Datum hashoidvector(PG_FUNCTION_ARGS);
-extern Datum hashint2vector(PG_FUNCTION_ARGS);
 extern Datum hashname(PG_FUNCTION_ARGS);
 extern Datum hashtext(PG_FUNCTION_ARGS);
 extern Datum hashvarlena(PG_FUNCTION_ARGS);
diff --git a/src/include/catalog/pg_amop.h b/src/include/catalog/pg_amop.h
index 15b6290..e4c3515 100644
--- a/src/include/catalog/pg_amop.h
+++ b/src/include/catalog/pg_amop.h
@@ -573,8 +573,6 @@ DATA(insert (	2040   1114 1114 1 s 2060 405 0 ));
 DATA(insert (	   16 16 1 s   91 405 0 ));
 /* bytea_ops */
 DATA(insert (	2223   17 17 1 s 1955 405 0 ));
-/* int2vector_ops */
-DATA(insert (	2224   22 22 1 s	386 405 0 ));
 /* xid_ops */
 DATA(insert (	2225   28 28 1 s	352 405 0 ));
 /* cid_ops */
diff --git a/src/include/catalog/pg_amproc.h 

Re: [HACKERS] int2vector and btree indexes

2016-10-11 Thread Tom Lane
Amit Langote  writes:
> I was wrong that the index *never* gets used.  It does in fact get used if
> the operator is an ordering search operator (<, <=, >, >=), in which case
> the query would use an array_ops operator (which is a btree operator class
> for type anyarray) and hence matches the index operator family.  I failed
> to mention in my original message that int2vector_ops is a hash operator
> class.  There is exactly one =(int2vector, int2vector) operator in the
> system of which there is no btree equivalent.

Hmm ... I kind of wonder why we have int2vectoreq or hashint2vector at
all, likewise the hash opclass based on them.  The code says that they
are needed to support catcache index columns, but the only columns of
this type are

regression=# select attrelid::regclass,attname from pg_attribute where atttypid 
= 'int2vector'::regtype;
  attrelid  |  attname  
+---
 pg_index   | indkey
 pg_index   | indoption
 pg_trigger | tgattr
(3 rows)

and those don't have indexes at all, let alone catcaches based on them.
So it looks to me like we could remove this infrastructure.  There is
value in being able to hash int2vectors during queries, for sure, but
we could let that be done by the anyarray hash opclass.

Having said that, int2vector is not meant as a user-facing type and so
I don't particularly care whether indexes built on it work conveniently.
But it looks to me like we've got some unnecessary code here.

regards, tom lane


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers


Re: [HACKERS] int2vector and btree indexes

2016-10-11 Thread Amit Langote
On 2016/10/11 15:58, Amit Langote wrote:
> If I create btree index on a int2vector column, it does not get used for
> queries because the query search always fails to match the index operator
> (family).
> 
> During index creation, GetDefaultOpClass() returns array_ops for a
> int2vector index column, because type int2vector is binary-coercible with
> anyarray (which is array_ops's input type).  Whereas queries involving
> int2vector columns would use a int2vector_ops operator.

I was wrong that the index *never* gets used.  It does in fact get used if
the operator is an ordering search operator (<, <=, >, >=), in which case
the query would use an array_ops operator (which is a btree operator class
for type anyarray) and hence matches the index operator family.  I failed
to mention in my original message that int2vector_ops is a hash operator
class.  There is exactly one =(int2vector, int2vector) operator in the
system of which there is no btree equivalent.

I guess there is not much to complaint about here after all.  Sorry about
the noise.

Thanks,
Amit




-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers