[CASSANDRA-1005] cassandra-cli doesn't work with system allowed column family names

• Type: Bug
• Status: Resolved
• Priority: Low
• Resolution: Fixed
• Fix Version/s: 0.6.2
• Labels: cassandra-cli

Environment

Windows XP 32 bit

• Severity: Low
• Assignee: Eric Evans
• Reporter: James Mello

Authors

Eric Evans

• Created: 19/Apr/10 20:34
• Updated: 16/Apr/19 09:33
• Resolved: 24/May/10 16:04

Description

Given the following definitions for columns:

<Keyspaces>

<Keyspace Name="NGram">

<KeysCachedFraction>0.01</KeysCachedFraction>

<ColumnFamily CompareWith="UTF8Type" Name="1GramR"/>

<ColumnFamily CompareWith="UTF8Type" Name="1GramL"/>

</Keyspaces>

The appropriate keyspaces are created an persisteted on startup. When executing a query or a set operation in the cassandra-cli, you end up with the following error:

******************************************************

cassandra> get NGram.1GramR['hte']

line 1:10 extraneous input '1' expecting Identifier

No such column family: GramR

******************************************************

Following the syntax of the grammer we can see the following:

setStmt
: K_SET columnFamilyExpr '=' value -> ^(NODE_THRIFT_SET columnFamilyExpr value)
;

...

columnFamilyExpr
: table DOT columnFamily '[' rowKey ']'
( '[' a+=columnOrSuperColumn ']'
('[' a+=columnOrSuperColumn ']')?
)?
-> ^(NODE_COLUMN_ACCESS table columnFamily rowKey ($a+)?)
;
...

// syntactic Elements
Identifier
: Letter ( Alnum | '_' )*
;

There is a mismatch on what is appropriate values for this in the system. So either the restriction needs to be lifted in the cli, or the system must have a way of honoring the names.