Recent changes to bugs

getGeneratedKeys()

Anonymous — Tue, 29 Jun 2004 16:56:51 -0000

The method getGeneratedKeys() of PreparedStatement
is only available since JDK 1.4, breaking the compatibility
with the JDK 1.3

Calls to this method are present at:

org.ldbc.or.Table (line 125)
org.ldbc.test.TestAutoIncrement (line 172)

loadAutoIncre() does not escapes metachars in self-requests.

Vladimirov Sergey — Sat, 03 Apr 2004 10:42:18 -0000

When loadAutoIncrementColumns() gets the column
names (Base.java, line 300), it does not escape table
names:
rs = meta.getColumns(null, null, tableName, null);

Will it be more correct? :
rs = meta.getColumns(null, null, quote(tableName), null);

getTables() requied too much time.

Vladimirov Sergey — Sat, 03 Apr 2004 10:37:46 -0000

It is bocose getTables() function load all tables name
from database.

But it need not for database, except MySQL. I think, it
will be good to add another boolen flag to Adapter class
(default value - false), to determine, need or not to use
such way of retriving tables. Or may be just pass to
standart metaDatabase function.

If it MySQL - use current function. If not - pass to JDBS
driver.

It is urgent, becose getTables() used by
loadAutoincrementColumn(), and this function used by
parser often.

Parser not check maximum column name and table nam lenght

Vladimirov Sergey — Thu, 11 Mar 2004 20:24:54 -0000

I get error from MS SQL server about too long column
name (36, but max is 30)

need count(1) instead count(*) for Oracle

Vladimirov Sergey — Thu, 11 Mar 2004 20:22:20 -0000

In Oracle all indexes ARE NOT USED if query contains
count(*), and all programmers should use COUNT(1)

But Count(1) is unexpected by LDBC.

adapter.properties

Vladimirov Sergey — Wed, 10 Mar 2004 17:58:53 -0000

adapter.properties not found in .jar (only in src.zip)

So it will be error on startup.

I put it file from src, but it will be more simple to put it
to .jar file at build stage.

mysql smallint and databasemetadata

Brian McAnulty — Wed, 29 Oct 2003 14:11:19 -0000

According to the fixed bug 659447 the smallint issue is
sorted, but I've just come across it again - had to
modify the MySQL adapter to include:

case Types.SMALLINT

just before

case Types.BIGINT

in convertDataType(...)

Parser problem with columns called timestamp

Brian McAnulty — Wed, 29 Oct 2003 11:27:57 -0000

Our DB currently has a column (right or wrong) called
'timestamp'. In a query we do something like :

WHERE timestamp BETWEEN ? AND ?

This doesn't parse as the parser seems to read the
'timestamp' token and assume its a SQL timestamp or
something. Tried prefixing timestamp with the table
name (where mytable.timestamp between ? and ?) but this
doesn't fix it - throws a different parser error to do
with expecting an identifier, not a timestamp.

Changing the column name isn't a particularly nice
option for us as there are support issues, so a more
intelligent parser would be great?!!!

Stack trace for the "identifer" exception:
line 1: expecting IDENTIFIER, found 'timestamp'
at org.ldbc.antlr.Parser.match(Parser.java:180)
at
org.ldbc.parser.SQLParser.identifier(SQLParser.java:2491)
at
org.ldbc.parser.SQLParser.column_name(SQLParser.java:2266)

at org.ldbc.parser.SQLParser.column(SQLParser.java:3558)
at
org.ldbc.parser.SQLParser.value_or_column(SQLParser.java:735)
at org.ldbc.parser.SQLParser.cond_end(SQLParser.java:656)
at org.ldbc.parser.SQLParser.cond_term(SQLParser.java:608)
at
org.ldbc.parser.SQLParser.cond_factor(SQLParser.java:526)
at org.ldbc.parser.SQLParser.cond_sum(SQLParser.java:477)
at org.ldbc.parser.SQLParser.cond_exp(SQLParser.java:448)
at org.ldbc.parser.SQLParser.cond_rel(SQLParser.java:204)
at org.ldbc.parser.SQLParser.cond_neg(SQLParser.java:165)
at org.ldbc.parser.SQLParser.cond_and(SQLParser.java:112)
at org.ldbc.parser.SQLParser.cond_or(SQLParser.java:73)
at org.ldbc.parser.SQLParser.condition(SQLParser.java:55)
at org.ldbc.parser.SQLParser.select(SQLParser.java:1641)
at
org.ldbc.parser.SQLParser.statement(SQLParser.java:1446)
at
org.ldbc.parser.Translator.translate(Translator.java:73)
at
org.ldbc.jdbc.jdbcConnection.translate(jdbcConnection.java:702)
at
org.ldbc.jdbc.jdbcPreparedStatement.<init>(jdbcPreparedStatement.java:1286)
at
org.ldbc.jdbc.jdbcConnection.prepareStatement(jdbcConnection.java:57)

and for the 'timestamp' exception:

line 1: expecting QUOTED_STRING, found 'BETWEEN'
at org.ldbc.antlr.Parser.match(Parser.java:180)
at
org.ldbc.parser.SQLParser.timestamp(SQLParser.java:3801)
at
org.ldbc.parser.SQLParser.value_no_null(SQLParser.java:3649)
at org.ldbc.parser.SQLParser.value(SQLParser.java:3585)
at
org.ldbc.parser.SQLParser.value_or_column(SQLParser.java:728)
at org.ldbc.parser.SQLParser.cond_end(SQLParser.java:656)
at org.ldbc.parser.SQLParser.cond_term(SQLParser.java:608)
at
org.ldbc.parser.SQLParser.cond_factor(SQLParser.java:526)
at org.ldbc.parser.SQLParser.cond_sum(SQLParser.java:477)
at org.ldbc.parser.SQLParser.cond_exp(SQLParser.java:448)
at org.ldbc.parser.SQLParser.cond_rel(SQLParser.java:204)
at org.ldbc.parser.SQLParser.cond_neg(SQLParser.java:165)
at org.ldbc.parser.SQLParser.cond_and(SQLParser.java:112)
at org.ldbc.parser.SQLParser.cond_or(SQLParser.java:73)
at org.ldbc.parser.SQLParser.condition(SQLParser.java:55)
at org.ldbc.parser.SQLParser.select(SQLParser.java:1641)
at
org.ldbc.parser.SQLParser.statement(SQLParser.java:1446)
at
org.ldbc.parser.Translator.translate(Translator.java:73)
at
org.ldbc.jdbc.jdbcConnection.translate(jdbcConnection.java:702)
at
org.ldbc.jdbc.jdbcPreparedStatement.<init>(jdbcPreparedStatement.java:1286)
at
org.ldbc.jdbc.jdbcConnection.prepareStatement(jdbcConnection.java:57)

Rgds
Brian

Configurable adapter list in Factory

Brian McAnulty — Wed, 29 Oct 2003 11:18:47 -0000

Currently the Factory contains a hardcoded list of
adapters, so if you want to write a new one you need to
change this class. Would be nice if this was a
configuration file - xml or simple properties file.

HSQLDB: autoincrement

Thomas Mueller — Tue, 09 Sep 2003 17:26:16 -0000

It can fail if multiple connections are used.