[
https://issues.apache.org/jira/browse/DERBY-3126?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12534182
]
kmarsden edited comment on DERBY-3126 at 10/11/07 5:53 PM:
-----------------------------------------------------------------
Attached is a patch for DERBY-3126. The patch is called derby-2073_diff.txt
The change changes NetStatementRequest.computeProtocolTypesAndLengths to always
look at the data scale and precision when sending decimal values to the server
rather than using the parameterMetaData values which are never set. For null
values it just uses scale 0 and precision 1. For non-null values I used the
pre-existing Utils.computeBigDecimalPrecision method to compute precision.
Getting the scale and precision from the data is the only way to make sure the
server gets the actual data value set by the user. There was some talk in the
old comments about scale and precision for output and inout parameters but as
far as I can tell the scale set with registerOutputParameter is not saved
anywhere, so cannot be used here. I don't think it makes sense use the output
scale for the input parameter as it might affect the value sent to the server.
There may be additional issues with output parameter and
registerOutputParameter, specifying a scale, since we don't seem to use the
scale anywhere. I filed DERBY-3125 for the lack of use of the scale parameter
in registerOutputParameter for both embedded and client.
I added a test to ParameterMappingTest which was only testing whole numbers for
decimal values.
I also noticed that setObject with big decimal and specified scale has a
similar issue, but it is not fixed by this patch. I filed DERBY-3127 for that.
tests passed
Kathey
was (Author: kmarsden):
Attached is a patch for DERBY-2073. The change changes
NetStatementRequest.computeProtocolTypesAndLengths to always look at the data
scale and precision when sending decimal values to the server rather than using
the parameterMetaData values which are never set. For null values it just uses
scale 0 and precision 1. For non-null values I used the pre-existing
Utils.computeBigDecimalPrecision method to compute precision.
Getting the scale and precision from the data is the only way to make sure the
server gets the actual data value set by the user. There was some talk in the
old comments about scale and precision for output and inout parameters but as
far as I can tell the scale set with registerOutputParameter is not saved
anywhere, so cannot be used here. I don't think it makes sense use the output
scale for the input parameter as it might affect the value sent to the server.
There may be additional issues with output parameter and
registerOutputParameter, specifying a scale, since we don't seem to use the
scale anywhere. I filed DERBY-3125 for the lack of use of the scale parameter
in registerOutputParameter for both embedded and client.
I added a test to ParameterMappingTest which was only testing whole numbers for
decimal values.
I also noticed that setObject with big decimal and specified scale has a
similar issue, but it is not fixed by this patch. I filed DERBY-3127 for that.
tests are running ...
Kathey
> Inserting BigDecimal value in PreparedStatement with setBigDecimal into
> double column loses fractional digits
> -------------------------------------------------------------------------------------------------------------
>
> Key: DERBY-3126
> URL: https://issues.apache.org/jira/browse/DERBY-3126
> Project: Derby
> Issue Type: Sub-task
> Components: JDBC, Network Client
> Reporter: Kathey Marsden
> Attachments: derby-2073_diff.txt, derby-2073_stat.txt
>
>
> Splitting this issue into two subtasks one for setBigDecimal and one for
> setObject as they take a different code path and I plan to submit separate
> patches. Reproduces with BigDecimalProblemTester attached to DERBY-2073
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
