Refactor implementation of {call ...} escape for better support in Statement, PreparedStatement and CallableStatement [JDBC296] #342
Comments
Modified by: @mrotteveelFix Version: Jaybird 2.3 [ 10440 ] |
Modified by: @mrotteveeldescription: Currently the {call ... } escape is supported for Statement and PreparedStatement as well. It should only be supported for CallableStatement. This will also reduce the complexity of parsing (with repeated parsing) as described in JDBC292. Implementation-wise the JDBCEscape parser should be changed to simply leave the {call ...} and {? = call ...} escapes untouched (but it should parse the escapes nested inside the call, leaving it upto Firebird to throw an excepion when it receives {call ...} untranslated. Then only the FBCallableStatement/FBEscapedCallParser implementation should translate the call escape to the appropriate escape. => Currently the {call ... } escape is supported for Statement and PreparedStatement as well. It should only be supported for CallableStatement. This will also reduce the complexity of parsing (with repeated parsing) as described in JDBC292. Implementation-wise the JDBCEscape parser should be changed to simply leave the {call ...} and {? = call ...} escapes untouched (but it should parse the escapes nested inside the call, leaving it upto Firebird to throw an excepion when it receives {call ...} untranslated. Then only the FBCallableStatement/FBEscapedCallParser implementation should translate the call escape to the appropriate Firebird specific call. |
Modified by: @mrotteveel |
|
Commented by: @mrotteveel Looking at section 6.4 of JDBC 4.1, it seems that the {call...} should also be supported on Statement and PreparedStatement. This means I have to find a different way of preventing double parsing when CallableStatement is used.... |
|
Commented by: @mrotteveel Sent a mail to jdbc-spec-discuss to ask whether the call-escape should be supported for Statement and PreparedStatement. |
|
Commented by: @mrotteveel Statement and PreparedStatement should support the {call } escape as well. For Statement only with literal parameters, for PreparedStatement only with IN parameters, and both don't need to support OUT parameters in the JDBC sense (but - of course - have to support ResultSet producing stored procedures). |
Modified by: @mrotteveeldescription: Currently the {call ... } escape is supported for Statement and PreparedStatement as well. It should only be supported for CallableStatement. This will also reduce the complexity of parsing (with repeated parsing) as described in JDBC292. Implementation-wise the JDBCEscape parser should be changed to simply leave the {call ...} and {? = call ...} escapes untouched (but it should parse the escapes nested inside the call, leaving it upto Firebird to throw an excepion when it receives {call ...} untranslated. Then only the FBCallableStatement/FBEscapedCallParser implementation should translate the call escape to the appropriate Firebird specific call. => The {call ...}-escape should be supported on Statement, PreparedStatement and CallableStatement (in various degrees), the current implementation is inefficient for CallableStatement (it parses the statement two or three times) and it looks like it might shuffle the parameter order in the case of {?= call ...} (not 100% sure), and support under Statement and PreparedStatement seems accidental (and there are no tests). Also the existing implementation considers parentheses around the parameter list to be optional; they are required by the syntax defined in section 13.4.4 of the JDBC spec (they are only optional if there are no parameters). See also http://mail.openjdk.java.net/pipermail/jdbc-spec-discuss/2013-January/000029.html ================================ Implementation-wise the JDBCEscape parser should be changed to simply leave the {call ...} and {? = call ...} escapes untouched (but it should parse the escapes nested inside the call, leaving it upto Firebird to throw an excepion when it receives {call ...} untranslated. Then only the FBCallableStatement/FBEscapedCallParser implementation should translate the call escape to the appropriate Firebird specific call. summary: Refactor support for {call ...} escape support so it can no longer be used from Statement and PreparedStatement. => Refactor support for {call ...} escape support for better support in Statement, PreparedStatement and CallableStatement |
Modified by: @mrotteveelsummary: Refactor support for {call ...} escape support for better support in Statement, PreparedStatement and CallableStatement => Refactor implementation of {call ...} escape for better support in Statement, PreparedStatement and CallableStatement |
Modified by: @mrotteveel |
Modified by: @mrotteveel |
|
Commented by: @mrotteveel See also JDBC402, one thing to check for is the validity of the current double parsing of the call escape, and whether the effect of FBProcedureCall.checkParameters() in FBEscapedParser.convertProcedureCall is actually the correct one. |
Modified by: @mrotteveel |
Submitted by: @mrotteveel
Is related to JDBC292
Relate to JDBC229
Relate to JDBC297
Relate to JDBC402
The {call ...}-escape should be supported on Statement, PreparedStatement and CallableStatement (in various degrees), the current implementation is inefficient for CallableStatement (it parses the statement two or three times) and it looks like it might shuffle the parameter order in the case of {?= call ...} (not 100% sure), and support under Statement and PreparedStatement seems accidental (and there are no tests).
Also the existing implementation considers parentheses around the parameter list to be optional; they are required by the syntax defined in section 13.4.4 of the JDBC spec (they are only optional if there are no parameters).
See also http://mail.openjdk.java.net/pipermail/jdbc-spec-discuss/2013-January/000029.html
================================
ORIGINAL TEXT:
Currently the {call ... } escape is supported for Statement and PreparedStatement as well. It should only be supported for CallableStatement. This will also reduce the complexity of parsing (with repeated parsing) as described in JDBC292.
Implementation-wise the JDBCEscape parser should be changed to simply leave the {call ...} and {? = call ...} escapes untouched (but it should parse the escapes nested inside the call, leaving it upto Firebird to throw an excepion when it receives {call ...} untranslated. Then only the FBCallableStatement/FBEscapedCallParser implementation should translate the call escape to the appropriate Firebird specific call.
The text was updated successfully, but these errors were encountered: