[
https://issues.apache.org/jira/browse/PHOENIX-6764?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17585168#comment-17585168
]
ASF GitHub Bot commented on PHOENIX-6764:
-----------------------------------------
stoty commented on PR #1490:
URL: https://github.com/apache/phoenix/pull/1490#issuecomment-1228096196
There two open issues:
1. The new litarals don't work with BINARY ARRAYS. I suspect that this is
not a new problem, so I'm not terribly concerned about that.
2. The way we convert binaries to strings is inconsitent:
On one hand, we have the Formatters that are used for VARBINARY (and now
BINARY) results, on the other we have the toStringLiteral() method in the
P*BBinary classes, which are used for Statement.toString(), and for
BINARY_ARRAY (which doesn't have an explicit Formatter set)
Before the patch, the formatter generated _'AABB'_, which the
toStringLiteral() methods returned _'[170, 187]'_
With my patch, the formatter is the same, only it's now also set for BINARY,
while previously it was only set for VARBINARY.
I've changed the toStringLiteral() formatter to return a valid binary string
literal, _X'AABB'_ , so now stmt.toString() generates valid SQL.
The complicate things, it seems that Oracle and Mysql simply returns
something like new String(byte[]) for getString(), so the Phoenix behaviour is
unique.
I don't think that this is show-stopper either, the patch doesn't make any
of those problems any worse.
> Implement Binary and Hexadecimal literals
> -----------------------------------------
>
> Key: PHOENIX-6764
> URL: https://issues.apache.org/jira/browse/PHOENIX-6764
> Project: Phoenix
> Issue Type: Bug
> Components: core
> Reporter: Istvan Toth
> Assignee: Istvan Toth
> Priority: Major
>
> Currently there is no sane way to specify arbitrary binary and varbinary
> values in the query string.
> They can be set as variables for preparedstatements, and if the length
> corresponds to an existing type, then some casting gymnastics can be used to
> work around the problem, but I have not found a way to write a query that
> upserts an arbitrary three byte value into a binary/varbinary.
> The SQL standard defines Binary and Hex literals in the form ofÂ
> B'01010101001...' and X'0102AAF5...'
> Implement this in the parser.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
