[
https://issues.apache.org/jira/browse/DERBY-5536?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13169620#comment-13169620
]
Dag H. Wanvik commented on DERBY-5536:
--------------------------------------
The bug also is present for getInt, getShort and getByte sinc ethey rely on
getLong. The error lies in am.Decimal#getLong, which contains this loop:
// compute the integer part only.
int leftOfDecimalPoint = length * 2 - 1 - scale;
long integer = 0;
if (leftOfDecimalPoint > 0) {
int i = 0;
for (; i < leftOfDecimalPoint / 2; i++) {
integer = integer * 10 + signum * ((buffer[offset + i] & 0xF0) >>>
4); // high nybble.
integer = integer * 10 + signum * (buffer[offset + i] & 0x0F);
// low nybble.
}
:
If the scale is high, say, 31, the accumulating value "integer" will wrap
around here with no warning.
> Client's ResultSet#getLong does not range check when converting from a
> DECIMAL column
> -------------------------------------------------------------------------------------
>
> Key: DERBY-5536
> URL: https://issues.apache.org/jira/browse/DERBY-5536
> Project: Derby
> Issue Type: Bug
> Components: JDBC, Network Client
> Reporter: Dag H. Wanvik
> Attachments: MissingRangeCheck.java
>
>
> Derby's DECIMAL can contain an integer of 31 digits. This can overflow a
> long. The embedded JDBC driver catches this, but the client does not and can
> yield a corrupt long, cf enclosed repro.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
