I found the following code in multirangetypes.c
> if (*ptr == '{')
> ptr++;
> else
> ereport(ERROR,
> (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION),
> errmsg("malformed multirange literal: \"%s\"",
> input_str),
> errdetail("Missing left bracket.")));
I'm not sure how much we (or people) are strcit on the distinction
between the $SUBJECT, isn't '{' a brace generally?
postgres=# select '[1,3]'::int4multirange;
ERROR: malformed multirange literal: "[1,3]"
LINE 1: select '[1,3]'::int4multirange;
^
DETAIL: Missing left bracket.
The distinction is significant there. It should at least be "Missing
left curly bracket." or "Missing left brace." (or left curly brace..?)
'{' is mentioned as "curly brackets" in comments of the soruce file.
It is mentioned as "brace" in regexp doc [1]. And.. uh.. I found the
world "curly braces" in the doc for conding conventions..
[1]: https://www.postgresql.org/docs/devel/functions-matching.html
regards.
--
Kyotaro Horiguchi
NTT Open Source Software Center
diff --git a/src/backend/utils/adt/multirangetypes.c b/src/backend/utils/adt/multirangetypes.c
index 0b81649779..fbcc27d072 100644
--- a/src/backend/utils/adt/multirangetypes.c
+++ b/src/backend/utils/adt/multirangetypes.c
@@ -146,7 +146,7 @@ multirange_in(PG_FUNCTION_ARGS)
(errcode(ERRCODE_INVALID_TEXT_REPRESENTATION),
errmsg("malformed multirange literal: \"%s\"",
input_str),
- errdetail("Missing left bracket.")));
+ errdetail("Missing left brace.")));
/* consume ranges */
parse_state = MULTIRANGE_BEFORE_RANGE;
@@ -282,7 +282,7 @@ multirange_in(PG_FUNCTION_ARGS)
(errcode(ERRCODE_INVALID_TEXT_REPRESENTATION),
errmsg("malformed multirange literal: \"%s\"",
input_str),
- errdetail("Junk after right bracket.")));
+ errdetail("Junk after right brace.")));
ret = make_multirange(mltrngtypoid, rangetyp, range_count, ranges);
PG_RETURN_MULTIRANGE_P(ret);
