On 2019-04-11 17:04, Jeevan Chalke wrote:
> The error is coming from get_collation_isdeterministic() when colloid
> passed is 0. I think like we do in get_collation_name(), we should
> return false here when such collation oid does not exist.
I'm not in favor of doing that. It would risk papering over errors of
omission at other call sites.
The root cause is that the same code match_pattern_prefix() is being
used for text and bytea, but bytea does not use collations, so having
the collation 0 is expected, and we shouldn't call
get_collation_isdeterministic() in that case.
Proposed patch attached.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
From acb1542a1f8ebee9c9d6d9322c64c849b2f23e15 Mon Sep 17 00:00:00 2001
From: Peter Eisentraut <pe...@eisentraut.org>
Date: Fri, 12 Apr 2019 09:49:10 +0200
Subject: [PATCH] Unbreak index optimization for LIKE on bytea
The same code is used to handle both text and bytea, but bytea is not
collation-aware, so we shouldn't call get_collation_isdeterministic()
in that case.
---
src/backend/utils/adt/like_support.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/src/backend/utils/adt/like_support.c
b/src/backend/utils/adt/like_support.c
index a65e63736c..7528c80f7c 100644
--- a/src/backend/utils/adt/like_support.c
+++ b/src/backend/utils/adt/like_support.c
@@ -267,8 +267,10 @@ match_pattern_prefix(Node *leftop,
* precise error messages.) (It should be possible to support at least
* Pattern_Prefix_Exact, but no point as along as the actual
* pattern-matching implementations don't support it.)
+ *
+ * expr_coll is not set for a non-collation-aware data type such as
bytea.
*/
- if (!get_collation_isdeterministic(expr_coll))
+ if (expr_coll && !get_collation_isdeterministic(expr_coll))
return NIL;
/*
--
2.21.0
